> ## Documentation Index > Fetch the complete documentation index at: https://docs.perplexity.ai/llms.txt > Use this file to discover all available pages before exploring further. # Pricing export const PRICING = { "_meta": { "purpose": "Single source of truth for the PricingCalculator widget. Every rate here is transcribed from the public docs; update prices HERE only.", "sources": { "agent.tools, agent.sandbox, search, sonar, embeddings": "/docs/getting-started/pricing", "agent.models": "/docs/agent-api/models (token rates are not on the pricing page)", "agent.presets": "model from the /docs/agent-api/presets Available Presets table. input/output tokens and per-run tool counts are median values from representative Agent API runs (editable in the widget), NOT billed values — actual cost is metered from each response's usage field.", "sonar": "/docs/getting-started/pricing Sonar API Pricing (Token Pricing + Request Pricing tables). citation/reasoning/searchQueries apply to Sonar Deep Research only. Pro Search (Sonar Pro variant) is intentionally not modeled here." }, "units": { "model input/output/cache": "$ per 1,000,000 tokens", "tools": "$ per invocation", "sandbox.session": "$ per session (<=20-min billing window)", "search.per1k": "$ per 1,000 requests", "sonar.input/output/citation/reasoning": "$ per 1,000,000 tokens", "sonar.request.{low,medium,high}": "$ per 1,000 requests (varies by search context size)", "sonar.searchQueries": "$ per 1,000 searches (Deep Research only)", "embeddings.rate": "$ per 1,000,000 tokens" }, "cacheEncoding": "number = flat $/1M cache-read rate; 'inputx0.1' = 90% off the active input rate; null = no cache pricing", "tieredEncoding": "input/output may be {low,high} objects; tierThreshold sets the per-model input-token switch point for the high tier" }, "agent": { "tools": { "web_search": 0.005, "fetch_url": 0.0005, "people_search": 0.005, "finance_search": 0.005 }, "sandbox": { "session": 0.03, "search": 0.005 }, "models": [{ "group": "Perplexity", "id": "perplexity/sonar", "input": 0.25, "output": 2.50, "cache": 0.0625 }, { "group": "Anthropic", "id": "anthropic/claude-opus-4-8", "input": 5, "output": 25, "cache": 0.50 }, { "group": "Anthropic", "id": "anthropic/claude-opus-4-7", "input": 5, "output": 25, "cache": 0.50 }, { "group": "Anthropic", "id": "anthropic/claude-opus-4-6", "input": 5, "output": 25, "cache": 0.50 }, { "group": "Anthropic", "id": "anthropic/claude-opus-4-5", "input": 5, "output": 25, "cache": 0.50 }, { "group": "Anthropic", "id": "anthropic/claude-sonnet-5", "input": 2, "output": 10, "cache": 0.20 }, { "group": "Anthropic", "id": "anthropic/claude-sonnet-4-6", "input": 3, "output": 15, "cache": 0.30 }, { "group": "Anthropic", "id": "anthropic/claude-sonnet-4-5", "input": 3, "output": 15, "cache": 0.30 }, { "group": "Anthropic", "id": "anthropic/claude-haiku-4-5", "input": 1, "output": 5, "cache": 0.10 }, { "group": "OpenAI", "id": "openai/gpt-5.6-sol", "input": { "low": 5.00, "high": 10.00 }, "output": { "low": 30.00, "high": 45.00 }, "cache": 0.50, "tierThreshold": 272000 }, { "group": "OpenAI", "id": "openai/gpt-5.6-terra", "input": { "low": 2.50, "high": 5.00 }, "output": { "low": 15.00, "high": 22.50 }, "cache": "inputx0.1", "tierThreshold": 272000 }, { "group": "OpenAI", "id": "openai/gpt-5.6-luna", "input": { "low": 1.00, "high": 2.00 }, "output": { "low": 6.00, "high": 9.00 }, "cache": 0.10, "tierThreshold": 272000 }, { "group": "OpenAI", "id": "openai/gpt-5.5", "input": { "low": 5.00, "high": 10.00 }, "output": { "low": 30.00, "high": 45.00 }, "cache": 0.50, "tierThreshold": 272000 }, { "group": "OpenAI", "id": "openai/gpt-5.4", "input": { "low": 2.50, "high": 5.00 }, "output": { "low": 15.00, "high": 22.50 }, "cache": 0.25, "tierThreshold": 272000 }, { "group": "OpenAI", "id": "openai/gpt-5.4-mini", "input": 0.75, "output": 4.50, "cache": 0.075 }, { "group": "OpenAI", "id": "openai/gpt-5.4-nano", "input": 0.20, "output": 1.25, "cache": 0.02 }, { "group": "OpenAI", "id": "openai/gpt-5.2", "input": 1.75, "output": 14, "cache": 0.175 }, { "group": "OpenAI", "id": "openai/gpt-5.1", "input": 1.25, "output": 10, "cache": 0.125 }, { "group": "OpenAI", "id": "openai/gpt-5", "input": 1.25, "output": 10, "cache": 0.125 }, { "group": "OpenAI", "id": "openai/gpt-5-mini", "input": 0.25, "output": 2, "cache": 0.025 }, { "group": "Google", "id": "google/gemini-3.1-pro-preview", "input": { "low": 2.00, "high": 4.00 }, "output": { "low": 12.00, "high": 18.00 }, "cache": "inputx0.1", "tierThreshold": 200000 }, { "group": "Google", "id": "google/gemini-3.1-flash-lite", "input": 0.25, "output": 1.50, "cache": "inputx0.1" }, { "group": "Google", "id": "google/gemini-3.1-flash-lite-preview", "input": 0.25, "output": 1.50, "cache": "inputx0.1" }, { "group": "Google", "id": "google/gemini-3.5-flash", "input": 1.50, "output": 9.00, "cache": 0.15 }, { "group": "Google", "id": "google/gemini-3-flash-preview", "input": 0.50, "output": 3.00, "cache": "inputx0.1" }, { "group": "xAI", "id": "xai/grok-4.5", "input": { "low": 2.00, "high": 4.00 }, "output": { "low": 6.00, "high": 12.00 }, "cache": 0.50, "tierThreshold": 200000 }, { "group": "xAI", "id": "xai/grok-4.3", "input": { "low": 1.25, "high": 2.50 }, "output": { "low": 2.50, "high": 5.00 }, "cache": 0.20, "tierThreshold": 200000 }, { "group": "xAI", "id": "xai/grok-4.20-reasoning", "input": { "low": 1.25, "high": 2.50 }, "output": { "low": 2.50, "high": 5.00 }, "cache": 0.20, "tierThreshold": 200000 }, { "group": "xAI", "id": "xai/grok-4.20-non-reasoning", "input": { "low": 1.25, "high": 2.50 }, "output": { "low": 2.50, "high": 5.00 }, "cache": 0.20, "tierThreshold": 200000 }, { "group": "xAI", "id": "xai/grok-4.20-multi-agent", "input": { "low": 1.25, "high": 2.50 }, "output": { "low": 2.50, "high": 5.00 }, "cache": 0.20, "tierThreshold": 200000 }, { "group": "Z.AI", "id": "perplexity/glm-5.2", "input": 1.40, "output": 4.40, "cache": 0.26 }, { "group": "Moonshot AI", "id": "perplexity/kimi-k2.7-code", "input": 0.95, "output": 4.00, "cache": 0.19 }, { "group": "NVIDIA", "id": "nvidia/nemotron-3-super-120b-a12b", "input": 0.25, "output": 2.50, "cache": null }], "presets": [{ "id": "fast", "model": "openai/gpt-5.4-mini", "input": 1000, "output": 500, "tools": { "web_search": 1, "fetch_url": 0 } }, { "id": "low", "model": "google/gemini-3-flash-preview", "input": 2000, "output": 1000, "tools": { "web_search": 1, "fetch_url": 1 } }, { "id": "medium", "model": "openai/gpt-5.6-luna", "input": 4000, "output": 1000, "tools": { "web_search": 2, "fetch_url": 2 } }, { "id": "high", "model": "openai/gpt-5.6-sol", "input": 4000, "output": 2000, "tools": { "web_search": 3, "fetch_url": 3 } }, { "id": "xhigh", "model": "openai/gpt-5.6-sol", "input": 8000, "output": 4000, "tools": { "web_search": 4, "fetch_url": 4, "sandbox_sessions": 1, "sandbox_searches": 2 } }] }, "search": { "per1k": 5.00 }, "sonar": { "models": [{ "id": "sonar", "label": "Sonar", "input": 1, "output": 1, "request": { "low": 5, "medium": 8, "high": 12 } }, { "id": "sonar-pro", "label": "Sonar Pro", "input": 3, "output": 15, "request": { "low": 6, "medium": 10, "high": 14 } }, { "id": "sonar-reasoning-pro", "label": "Sonar Reasoning Pro", "input": 2, "output": 8, "request": { "low": 6, "medium": 10, "high": 14 } }, { "id": "sonar-deep-research", "label": "Sonar Deep Research", "input": 2, "output": 8, "citation": 2, "reasoning": 3, "searchQueries": 5 }] }, "embeddings": [{ "id": "pplx-embed-v1-0.6b", "dims": 1024, "rate": 0.004 }, { "id": "pplx-embed-v1-4b", "dims": 2560, "rate": 0.03 }, { "id": "pplx-embed-context-v1-0.6b", "dims": 1024, "rate": 0.008 }, { "id": "pplx-embed-context-v1-4b", "dims": 2560, "rate": 0.05 }] }; export const PricingCalculator = ({product, data} = {}) => { const PRICING = data; const dataValid = !!(PRICING && PRICING.agent && PRICING.agent.tools && PRICING.agent.sandbox && Array.isArray(PRICING.agent.models) && PRICING.agent.models.length && Array.isArray(PRICING.agent.presets) && PRICING.agent.presets.length && PRICING.search && typeof PRICING.search.per1k === 'number' && PRICING.sonar && Array.isArray(PRICING.sonar.models) && PRICING.sonar.models.length && Array.isArray(PRICING.embeddings) && PRICING.embeddings.length); if (!dataValid) { return

Pricing data unavailable.

; } const TOOL_PRICE = PRICING.agent.tools; const SANDBOX_SESSION = PRICING.agent.sandbox.session; const SANDBOX_SEARCH = PRICING.agent.sandbox.search; const AGENT_MODELS = PRICING.agent.models; const AGENT_PRESETS = PRICING.agent.presets; const EMB_MODELS = PRICING.embeddings || []; const SEARCH_PER_1K = PRICING.search ? PRICING.search.per1k : undefined; const SONAR_MODELS = PRICING.sonar && PRICING.sonar.models || []; const PRODUCTS = ['search', 'agent', 'sonar', 'embeddings']; const PRODUCT_META = { agent: { label: 'Agent API', accent: 'var(--cb-agent-api-fg)' }, search: { label: 'Search API', accent: 'var(--cb-search-api-fg)' }, sonar: { label: 'Sonar API', accent: 'var(--cb-sonar-api-fg)' }, embeddings: { label: 'Embeddings API', accent: 'var(--cb-embeddings-api-fg)' } }; const isSingleProduct = !!product && PRODUCTS.includes(product); const [activeProduct, setActiveProduct] = useState(isSingleProduct ? product : 'search'); const DEFAULT_PRESET_ID = 'fast'; const DEFAULT_PRESET = AGENT_PRESETS.find(p => p.id === DEFAULT_PRESET_ID) || AGENT_PRESETS[0]; const DP_TOOLS = DEFAULT_PRESET.tools || ({}); const [agentModelId, setAgentModelId] = useState(AGENT_MODELS.some(m => m.id === 'perplexity/sonar') ? 'perplexity/sonar' : DEFAULT_PRESET.model); const [agentInput, setAgentInput] = useState(DEFAULT_PRESET.input); const [agentOutput, setAgentOutput] = useState(DEFAULT_PRESET.output); const [agentWebSearch, setAgentWebSearch] = useState(DP_TOOLS.web_search || 0); const [agentFetchUrl, setAgentFetchUrl] = useState(DP_TOOLS.fetch_url || 0); const [agentPeople, setAgentPeople] = useState(DP_TOOLS.people_search || 0); const [agentFinance, setAgentFinance] = useState(DP_TOOLS.finance_search || 0); const [agentSandboxSessions, setAgentSandboxSessions] = useState(DP_TOOLS.sandbox_sessions || 0); const [agentSandboxSearches, setAgentSandboxSearches] = useState(DP_TOOLS.sandbox_searches || 0); const [agentRunsPerMonth, setAgentRunsPerMonth] = useState(1000); const [searchRequests, setSearchRequests] = useState(1000); const [sonarModelId, setSonarModelId] = useState('sonar'); const [sonarInput, setSonarInput] = useState(1000); const [sonarOutput, setSonarOutput] = useState(1000); const [sonarContext, setSonarContext] = useState('low'); const [sonarRequests, setSonarRequests] = useState(1000); const [sonarCitation, setSonarCitation] = useState(20000); const [sonarReasoning, setSonarReasoning] = useState(74000); const [sonarSearches, setSonarSearches] = useState(18); const [embModelId, setEmbModelId] = useState('pplx-embed-v1-0.6b'); const [embTokens, setEmbTokens] = useState(100000); const [volume, setVolume] = useState(1000); const [usageMode, setUsageMode] = useState('total'); const [agentTotalInputM, setAgentTotalInputM] = useState(50); const [agentTotalOutputM, setAgentTotalOutputM] = useState(10); const [agentTotalWebSearch, setAgentTotalWebSearch] = useState(0); const [agentTotalFetchUrl, setAgentTotalFetchUrl] = useState(0); const [agentTotalPeople, setAgentTotalPeople] = useState(0); const [agentTotalFinance, setAgentTotalFinance] = useState(0); const [agentTotalSandboxSessions, setAgentTotalSandboxSessions] = useState(0); const [agentTotalSandboxSearches, setAgentTotalSandboxSearches] = useState(0); const [sonarTotalInputM, setSonarTotalInputM] = useState(50); const [sonarTotalOutputM, setSonarTotalOutputM] = useState(10); const [sonarTotalRequests, setSonarTotalRequests] = useState(1000); const [sonarTotalCitationM, setSonarTotalCitationM] = useState(20); const [sonarTotalReasoningM, setSonarTotalReasoningM] = useState(74); const [sonarTotalSearches, setSonarTotalSearches] = useState(18000); const [toolsOpen, setToolsOpen] = useState(false); const [modelFilter, setModelFilter] = useState(''); const [modelSort, setModelSort] = useState('cost'); const [modelProvider, setModelProvider] = useState('all'); const [showBreakdown, setShowBreakdown] = useState(false); const [reduceMotion, setReduceMotion] = useState(false); const [liveText, setLiveText] = useState(''); const [openInfo, setOpenInfo] = useState(null); const [infoPlace, setInfoPlace] = useState({ x: 'left', y: 'down' }); const infoRefs = useRef({}); const [openListbox, setOpenListbox] = useState(null); const [activeOptIdx, setActiveOptIdx] = useState(-1); const listboxRefs = useRef({}); const typeaheadRef = useRef({ buf: '', t: 0 }); const didSyncUrl = useRef(false); const modelListRef = useRef(null); const [copied, setCopied] = useState(false); const [showFormula, setShowFormula] = useState(false); useEffect(() => { if (typeof window === 'undefined' || !window.matchMedia) return undefined; const mq = window.matchMedia('(prefers-reduced-motion: reduce)'); const update = () => setReduceMotion(mq.matches); update(); if (mq.addEventListener) mq.addEventListener('change', update); else if (mq.addListener) mq.addListener(update); return () => { if (mq.removeEventListener) mq.removeEventListener('change', update); else if (mq.removeListener) mq.removeListener(update); }; }, []); useEffect(() => { if (isSingleProduct || typeof window === 'undefined') return; const params = new URLSearchParams(window.location.search); const fromUrl = params.get('calc'); if (fromUrl && PRODUCTS.includes(fromUrl)) setActiveProduct(fromUrl); }, []); useEffect(() => { if (isSingleProduct || typeof window === 'undefined') return; if (!didSyncUrl.current) { didSyncUrl.current = true; return; } const params = new URLSearchParams(window.location.search); if (params.get('calc') === activeProduct) return; params.set('calc', activeProduct); const qs = params.toString(); const url = window.location.pathname + (qs ? `?${qs}` : '') + (window.location.hash || ''); window.history.replaceState(null, '', url); }, [activeProduct]); useEffect(() => { const container = modelListRef.current; if (!container) return; const sel = container.querySelector('[role="radio"][aria-checked="true"]'); if (!sel) return; const c = container.getBoundingClientRect(); const s = sel.getBoundingClientRect(); if (s.top < c.top) container.scrollTop += s.top - c.top - 8; else if (s.bottom > c.bottom) container.scrollTop += s.bottom - c.bottom + 8; }, [agentModelId]); useEffect(() => { if (num(agentSandboxSessions) === 0 && num(agentSandboxSearches) !== 0) setAgentSandboxSearches(0); }, [agentSandboxSessions]); useEffect(() => { if (num(agentTotalSandboxSessions) === 0 && num(agentTotalSandboxSearches) !== 0) setAgentTotalSandboxSearches(0); }, [agentTotalSandboxSessions]); useEffect(() => { if (!openListbox || typeof document === 'undefined') return undefined; const onClick = e => { const el = listboxRefs.current[openListbox]; if (el && !el.contains(e.target)) setOpenListbox(null); }; const onKey = e => { if (e.key === 'Escape') setOpenListbox(null); }; document.addEventListener('mousedown', onClick); document.addEventListener('keydown', onKey); return () => { document.removeEventListener('mousedown', onClick); document.removeEventListener('keydown', onKey); }; }, [openListbox]); useEffect(() => { if (!openInfo || typeof document === 'undefined') return undefined; const onClick = e => { const el = infoRefs.current[openInfo]; if (el && !el.contains(e.target)) setOpenInfo(null); }; const onKey = e => { if (e.key === 'Escape') setOpenInfo(null); }; document.addEventListener('mousedown', onClick); document.addEventListener('keydown', onKey); return () => { document.removeEventListener('mousedown', onClick); document.removeEventListener('keydown', onKey); }; }, [openInfo]); const num = x => Math.max(0, Number(x) || 0); const intVal = x => { const n = Math.floor(Number(x)); return isFinite(n) && n > 0 ? n : 0; }; const floatVal = (x, dp = 3) => { const n = Number(x); if (!isFinite(n) || n <= 0) return 0; const f = Math.pow(10, dp); return Math.round(n * f) / f; }; const sanitizeDecimal = raw => { const s = String(raw).replace(/[^\d.]/g, ''); const i = s.indexOf('.'); return i < 0 ? s : s.slice(0, i + 1) + s.slice(i + 1).replace(/\./g, ''); }; const tokenCost = (n, ratePer1M) => num(n) / 1e6 * ratePer1M; const per1k = (n, feePer1K) => num(n) / 1000 * feePer1K; const fmt = v => { if (!isFinite(v) || v === 0) return '$0.00'; if (v >= 10000) { return '$' + Math.round(v).toLocaleString('en-US'); } if (v >= 1) { const hasCents = Math.round(v * 100) % 100 !== 0; return '$' + v.toLocaleString('en-US', { minimumFractionDigits: hasCents ? 2 : 0, maximumFractionDigits: 2 }); } if (v < 0.001) return '<$0.001'; let s = v.toFixed(3).replace(/0+$/, ''); if ((s.split('.')[1] || '').length < 2) s = v.toFixed(2); return '$' + s; }; const fmtRate = v => { if (!isFinite(v)) return '$0'; if (Number.isInteger(v)) return '$' + v; let s = v.toFixed(4).replace(/0+$/, ''); if ((s.split('.')[1] || '').length < 2) s = v.toFixed(2); return '$' + s; }; const q = x => num(x).toLocaleString('en-US'); const qRun = x => { const n = num(x); return Number.isInteger(n) ? n.toLocaleString('en-US') : Number(n.toFixed(2)).toString(); }; const qM = x => { const n = num(x); return Number.isInteger(n) ? n.toLocaleString('en-US') : Number(n.toFixed(3)).toString(); }; const agentModel = AGENT_MODELS.find(m => m.id === agentModelId) || AGENT_MODELS[0]; const DEFAULT_TIER_THRESHOLD = 200000; const isTieredModel = m => m.input && typeof m.input === 'object' || m.output && typeof m.output === 'object'; const tierThresholdFor = m => { const threshold = num(m && m.tierThreshold); return threshold > 0 ? threshold : DEFAULT_TIER_THRESHOLD; }; const formatTokenThreshold = threshold => { const n = num(threshold); return n >= 1000 && n % 1000 === 0 ? `${n / 1000}k` : n.toLocaleString('en-US'); }; const rateFor = m => { const tierThreshold = tierThresholdFor(m); const highTier = usageMode !== 'total' && num(agentInput) > tierThreshold; const pick = v => v && typeof v === 'object' ? v[highTier ? 'high' : 'low'] : v; return { inRate: pick(m.input), outRate: pick(m.output), highTier, tierThreshold, tiered: isTieredModel(m) }; }; const computeAgent = () => { const m = agentModel; const {inRate, outRate, tiered, highTier, tierThreshold} = rateFor(m); const tier = highTier ? 'high' : 'low'; if (usageMode === 'total') { const inputCost = num(agentTotalInputM) * inRate; const outputCost = num(agentTotalOutputM) * outRate; const toolWeb = num(agentTotalWebSearch) * TOOL_PRICE.web_search; const toolFetch = num(agentTotalFetchUrl) * TOOL_PRICE.fetch_url; const toolPeople = num(agentTotalPeople) * TOOL_PRICE.people_search; const toolFinance = num(agentTotalFinance) * TOOL_PRICE.finance_search; const toolsCost = toolWeb + toolFetch + toolPeople + toolFinance; const sess = num(agentTotalSandboxSessions); const srch = sess > 0 ? num(agentTotalSandboxSearches) : 0; const sandboxCost = sess * SANDBOX_SESSION + srch * SANDBOX_SEARCH; const total = inputCost + outputCost + toolsCost + sandboxCost; return { total, tiered, tier, tierThreshold, inputCost, outputCost, toolsCost, sandboxCost, tools: { web: toolWeb, fetch: toolFetch, people: toolPeople, finance: toolFinance } }; } const inputCost = tokenCost(num(agentInput), inRate); const outputCost = tokenCost(agentOutput, outRate); const toolWeb = num(agentWebSearch) * TOOL_PRICE.web_search; const toolFetch = num(agentFetchUrl) * TOOL_PRICE.fetch_url; const toolPeople = num(agentPeople) * TOOL_PRICE.people_search; const toolFinance = num(agentFinance) * TOOL_PRICE.finance_search; const toolsCost = toolWeb + toolFetch + toolPeople + toolFinance; const sandboxCost = num(agentSandboxSessions) * SANDBOX_SESSION + num(agentSandboxSearches) * SANDBOX_SEARCH; const total = inputCost + outputCost + toolsCost + sandboxCost; return { total, tiered, tier, tierThreshold, inputCost, outputCost, toolsCost, sandboxCost, tools: { web: toolWeb, fetch: toolFetch, people: toolPeople, finance: toolFinance } }; }; const computeSearch = () => ({ total: per1k(searchRequests, SEARCH_PER_1K) }); const embModel = EMB_MODELS.find(m => m.id === embModelId) || EMB_MODELS[0]; const computeEmbeddings = () => ({ total: tokenCost(embTokens, embModel.rate) }); const sonarModel = SONAR_MODELS.find(m => m.id === sonarModelId) || SONAR_MODELS[0]; const sonarIsDeepResearch = !!(sonarModel && sonarModel.searchQueries != null); const computeSonar = () => { const m = sonarModel; if (usageMode === 'total') { const inputCost = num(sonarTotalInputM) * m.input; const outputCost = num(sonarTotalOutputM) * m.output; if (m.searchQueries != null) { const citationCost = num(sonarTotalCitationM) * m.citation; const reasoningCost = num(sonarTotalReasoningM) * m.reasoning; const searchesCost = per1k(sonarTotalSearches, m.searchQueries); const total = inputCost + outputCost + citationCost + reasoningCost + searchesCost; return { total, inputCost, outputCost, citationCost, reasoningCost, searchesCost, requestFee: 0, deepResearch: true }; } const requestFee = per1k(sonarTotalRequests, m.request[sonarContext]); const total = inputCost + outputCost + requestFee; return { total, inputCost, outputCost, requestFee, deepResearch: false }; } const inputCost = tokenCost(sonarInput, m.input); const outputCost = tokenCost(sonarOutput, m.output); if (m.searchQueries != null) { const citationCost = tokenCost(sonarCitation, m.citation); const reasoningCost = tokenCost(sonarReasoning, m.reasoning); const searchesCost = per1k(sonarSearches, m.searchQueries); const total = inputCost + outputCost + citationCost + reasoningCost + searchesCost; return { total, inputCost, outputCost, citationCost, reasoningCost, searchesCost, requestFee: 0, deepResearch: true }; } const requestFee = per1k(1, m.request[sonarContext]); const total = inputCost + outputCost + requestFee; return { total, inputCost, outputCost, requestFee, deepResearch: false }; }; const result = activeProduct === 'agent' ? computeAgent() : activeProduct === 'search' ? computeSearch() : activeProduct === 'sonar' ? computeSonar() : computeEmbeddings(); const isAgent = activeProduct === 'agent'; const isSearch = activeProduct === 'search'; const isSonar = activeProduct === 'sonar'; const isTotalMode = (isAgent || isSonar) && usageMode === 'total'; const volN = isAgent ? num(agentRunsPerMonth) : isSearch ? num(searchRequests) : isSonar ? num(sonarRequests) : num(volume); const unitCost = isSearch ? SEARCH_PER_1K / 1000 : result.total || 0; const projected = isSearch || isTotalMode ? result.total : unitCost * volN; const displayProjected = fmt(projected); useEffect(() => { const handle = setTimeout(() => { const meta = PRODUCT_META[activeProduct]; let suffix = ''; if (isTotalMode) suffix = ' for your total usage'; else if (isAgent) suffix = ` for ${q(agentRunsPerMonth)} API calls`; else if (isSearch) suffix = ` for ${q(searchRequests)} requests`; else if (isSonar) suffix = ` for ${q(sonarRequests)} API calls`; else if (num(volume) > 1) suffix = ` for ${q(volume)} requests`; setLiveText(`${meta.label} estimate: ${displayProjected}${suffix}`); }, 400); return () => clearTimeout(handle); }, [displayProjected, activeProduct, volume, searchRequests, sonarRequests, agentRunsPerMonth, isAgent, isSearch, isSonar, isTotalMode]); const fg = { color: 'var(--color-foreground)' }; const muted = { color: 'var(--color-muted-foreground)' }; const borderStyle = { borderColor: 'var(--color-border)' }; const segSelected = 'bg-[#1215160f] dark:bg-[#f7f7f81a]'; const motionCls = reduceMotion ? '' : 'transition-all duration-200'; const fieldBase = 'rounded-[8px] border bg-transparent px-3 py-1.5 text-sm tabular-nums focus:outline-none focus-visible:ring-2 focus-visible:ring-[color:var(--calc-ring)]'; const chipBtn = 'rounded-full border px-2.5 py-1 text-xs tabular-nums hover:bg-[var(--cb-hover)] focus:outline-none focus-visible:ring-2 focus-visible:ring-[color:var(--calc-ring)]'; const renderNumberField = (label, value, setValue, id, opts = {}) => { const {helper, info, slider, narrow, max, min, decimals} = opts; const coerce = raw => { let n = decimals ? floatVal(raw, decimals) : intVal(raw); if (typeof min === 'number') n = Math.max(min, n); if (typeof max === 'number') n = Math.min(max, n); return n; }; const inputEl = setValue(decimals ? sanitizeDecimal(e.target.value) : e.target.value.replace(/[^\d.]/g, '').split('.')[0])} onBlur={e => setValue(coerce(e.target.value))} aria-describedby={helper ? `${id}-help` : undefined} className={`${fieldBase} ${narrow ? 'w-20 shrink-0 sm:w-24' : 'w-full'}`} style={borderStyle} />; return
{slider ? <> {}
{info}
{inputEl}
{slider} : <>
{info}
{inputEl} } {helper &&

{helper}

}
; }; const ladder125 = (lo, hi) => { const out = []; const mag = Math.floor(Math.log10(Math.max(lo, 1))); for (let e = mag; e <= Math.ceil(Math.log10(hi)) + 1; e++) { for (const m of [1, 2, 5]) { const val = m * Math.pow(10, e); if (val >= lo && val <= hi) out.push(val); } } if (!out.length || out[0] > lo) out.unshift(lo); if (out[out.length - 1] < hi) out.push(hi); return out; }; const snapTo125 = (v, lo, hi) => { const rungs = ladder125(lo, hi); return rungs.reduce((best, r) => Math.abs(Math.log10(r) - Math.log10(Math.max(v, lo))) < Math.abs(Math.log10(best) - Math.log10(Math.max(v, lo))) ? r : best, rungs[0]); }; const nextRung125 = (v, lo, hi, dir) => { const rungs = ladder125(lo, hi); const cur = snapTo125(v, lo, hi); const i = rungs.indexOf(cur); const ni = Math.min(rungs.length - 1, Math.max(0, i + dir)); return rungs[ni]; }; const renderSliderBar = (value, setValue, id, opts = {}) => { const {scale = 'linear', min = 0, max: maxProp = 100, step = 1, labelId, valueText, logMin} = opts; const v = num(value); const max = Math.max(maxProp, v); const isLog = scale === 'log'; const lo = isLog ? typeof logMin === 'number' && logMin > 0 ? logMin : Math.max(1, min) : min; const clamp = x => Math.min(max, Math.max(min, x)); const lg = x => Math.log10(Math.max(x, lo)); const posPct = x => { const cx = clamp(x); return isLog ? (lg(cx) - lg(lo)) / (lg(max) - lg(lo)) * 100 : (cx - min) / (max - min) * 100; }; const valueAtPos = pct => { const p = Math.min(1, Math.max(0, pct)); if (isLog) return snapTo125(Math.pow(10, lg(lo) + p * (lg(max) - lg(lo))), lo, max); return clamp(Math.round((min + p * (max - min)) / step) * step); }; const fillPct = posPct(v); const onPointer = (e, el) => { if (el) { const r = el.getBoundingClientRect(); setValue(valueAtPos((e.clientX - r.left) / r.width)); } }; const startDrag = (e, track) => { onPointer(e, track); const move = ev => onPointer(ev, track); const up = () => { window.removeEventListener('pointermove', move); window.removeEventListener('pointerup', up); }; window.addEventListener('pointermove', move); window.addEventListener('pointerup', up); }; const onThumbKeyDown = e => { let nv = null; const up1 = () => isLog ? nextRung125(v, lo, max, 1) : clamp(v + step); const dn1 = () => isLog ? nextRung125(v, lo, max, -1) : clamp(v - step); if (e.key === 'ArrowRight' || e.key === 'ArrowUp') nv = up1(); else if (e.key === 'ArrowLeft' || e.key === 'ArrowDown') nv = dn1(); else if (e.key === 'Home') nv = min; else if (e.key === 'End') nv = max; else if (e.key === 'PageUp') nv = isLog ? snapTo125(clamp(v * 10), lo, max) : clamp(v + 10 * step); else if (e.key === 'PageDown') nv = isLog ? snapTo125(clamp(v / 10), lo, max) : clamp(v - 10 * step); if (nv !== null) { e.preventDefault(); setValue(nv); } }; return
startDrag(e, e.currentTarget)}>
{ e.stopPropagation(); startDrag(e, e.currentTarget.parentElement); }}>
; }; const renderListbox = (id, value, setValue, options, {grouped = false, renderOption} = {}) => { const open = openListbox === id; const selected = options.find(o => o.id === value) || options[0]; const optLabel = renderOption || (o => o.id); const activeId = open && activeOptIdx >= 0 && options[activeOptIdx] ? `${id}-opt-${activeOptIdx}` : undefined; const openWith = idx => { setActiveOptIdx(idx); setOpenListbox(id); }; const closeRefocus = () => { setOpenListbox(null); if (typeof document !== 'undefined') { const t = document.getElementById(`${id}-trigger`); if (t) t.focus(); } }; const choose = idx => { if (options[idx]) { setValue(options[idx].id); closeRefocus(); } }; const moveActive = dir => setActiveOptIdx((activeOptIdx + dir + options.length) % options.length); const onTriggerKeyDown = e => { if (e.key === 'Enter' || e.key === ' ' || e.key === 'ArrowDown' || e.key === 'ArrowUp') { e.preventDefault(); openWith(Math.max(0, options.findIndex(o => o.id === value))); } }; const onListKeyDown = e => { if (e.key === 'ArrowDown') { e.preventDefault(); moveActive(1); } else if (e.key === 'ArrowUp') { e.preventDefault(); moveActive(-1); } else if (e.key === 'Home') { e.preventDefault(); setActiveOptIdx(0); } else if (e.key === 'End') { e.preventDefault(); setActiveOptIdx(options.length - 1); } else if (e.key === 'Enter' || e.key === ' ') { e.preventDefault(); choose(activeOptIdx); } else if (e.key === 'Escape') { e.preventDefault(); closeRefocus(); } else if (e.key === 'Tab') { setOpenListbox(null); } else if (e.key.length === 1 && (/\S/).test(e.key)) { const now = Date.now(); const ta = typeaheadRef.current; ta.buf = now - ta.t > 500 ? e.key.toLowerCase() : ta.buf + e.key.toLowerCase(); ta.t = now; const shortId = o => o.id.split('/').pop(); const hit = options.findIndex(o => shortId(o).toLowerCase().startsWith(ta.buf) || o.id.toLowerCase().startsWith(ta.buf)); if (hit >= 0) setActiveOptIdx(hit); } }; const rows = []; let lastGroup = null; options.forEach((opt, i) => { if (grouped && opt.group !== lastGroup) { const firstHeader = lastGroup === null; rows.push(
  • {opt.group}
    • ); lastGroup = opt.group; } const isSel = opt.id === value; const isActive = i === activeOptIdx; rows.push(
  • setActiveOptIdx(i)} onClick={() => choose(i)} className={`flex cursor-pointer items-center justify-between gap-2 rounded-[6px] px-2 py-1.5 text-sm ${isSel ? 'bg-[#12151614] dark:bg-[#f7f7f824] font-medium' : isActive ? 'bg-[var(--cb-hover)]' : ''} ${isActive ? 'ring-1 ring-inset ring-[color:var(--calc-ring)]' : ''}`} style={fg}> {optLabel(opt)} {isSel && }
    • ); }); return
    { listboxRefs.current[id] = el; }}> {open &&
      { if (el) el.focus(); }} onKeyDown={onListKeyDown} className="absolute z-50 mt-2 max-h-[320px] w-full overflow-y-auto rounded-lg border p-1 shadow-lg focus:outline-none" style={{ backgroundColor: 'var(--cb-popover-bg)', borderColor: 'var(--cb-popover-border)' }}> {rows}
    }
    ; }; const presetLabel = pid => ({ 'fast': 'Quick lookup', 'low': 'Everyday research', 'medium': 'Multi-source research', 'high': 'Exhaustive analysis', 'xhigh': 'Open-ended agentic work' })[pid] || pid; const applyTier = tier => { const p = AGENT_PRESETS.find(x => x.id === tier); if (!p) return; const t = p.tools || ({}); setAgentModelId(p.model); setAgentInput(p.input); setAgentOutput(p.output); setAgentWebSearch(t.web_search || 0); setAgentFetchUrl(t.fetch_url || 0); setAgentPeople(t.people_search || 0); setAgentFinance(t.finance_search || 0); setAgentSandboxSessions(t.sandbox_sessions || 0); setAgentSandboxSearches(t.sandbox_searches || 0); }; const agentRates = (() => { const {inRate, outRate} = rateFor(agentModel); return { inR: inRate, outR: outRate }; })(); const agentEstimateText = () => { const r = result; if (usageMode === 'total') { return ['Perplexity Agent API cost estimate (total usage)', `Model: ${agentModelId}`, `Input tokens: ${qM(agentTotalInputM)}M`, `Output tokens: ${qM(agentTotalOutputM)}M`, `Tools (total): web_search ${q(agentTotalWebSearch)}, fetch_url ${q(agentTotalFetchUrl)}, people_search ${q(agentTotalPeople)}, finance_search ${q(agentTotalFinance)}`, `Sandbox (total): ${q(agentTotalSandboxSessions)} sessions, ${q(agentTotalSandboxSearches)} searches`, `Estimated cost: ${fmt(r.total)}`, 'Estimate only. Final cost is metered from each response usage field.'].join('\n'); } const lines = ['Perplexity Agent API cost estimate', `Model: ${agentModelId}`, `API calls: ${q(agentRunsPerMonth)}`, `Input tokens: ${q(agentInput)}`, `Output tokens: ${q(agentOutput)}`, `Tools / call: web_search ${qRun(agentWebSearch)}, fetch_url ${qRun(agentFetchUrl)}, people_search ${qRun(agentPeople)}, finance_search ${qRun(agentFinance)}`, `Sandbox / call: ${qRun(agentSandboxSessions)} sessions, ${qRun(agentSandboxSearches)} searches`, `Per-call cost: ${fmt(r.total)}`, `Estimated cost: ${fmt(r.total * num(agentRunsPerMonth))}`, 'Estimate only. Final cost is metered from each response usage field.']; return lines.join('\n'); }; const copyEstimate = () => { const text = estimateText(); if (typeof navigator !== 'undefined' && navigator.clipboard && navigator.clipboard.writeText) { navigator.clipboard.writeText(text).then(() => { setCopied(true); setTimeout(() => setCopied(false), 2000); }).catch(() => {}); } else if (typeof document !== 'undefined') { try { const ta = document.createElement('textarea'); ta.value = text; ta.style.position = 'fixed'; ta.style.opacity = '0'; document.body.appendChild(ta); ta.select(); document.execCommand('copy'); document.body.removeChild(ta); setCopied(true); setTimeout(() => setCopied(false), 2000); } catch (e) {} } }; const agentFormulaText = () => { const {inR, outR} = agentRates; if (usageMode === 'total') { const parts = [`${qM(agentTotalInputM)}M × ${fmtRate(inR)}/1M`, `${qM(agentTotalOutputM)}M × ${fmtRate(outR)}/1M`]; const bits = []; if (num(agentTotalSandboxSessions) > 0) bits.push(`sandbox ${q(agentTotalSandboxSessions)}×${fmtRate(SANDBOX_SESSION)}`); if (num(agentTotalSandboxSearches) > 0) bits.push(`sandbox-search ${q(agentTotalSandboxSearches)}×${fmtRate(SANDBOX_SEARCH)}`); if (num(agentTotalWebSearch) > 0) bits.push(`web ${q(agentTotalWebSearch)}×${fmtRate(TOOL_PRICE.web_search)}`); if (num(agentTotalFetchUrl) > 0) bits.push(`fetch ${q(agentTotalFetchUrl)}×${fmtRate(TOOL_PRICE.fetch_url)}`); if (num(agentTotalPeople) > 0) bits.push(`people ${q(agentTotalPeople)}×${fmtRate(TOOL_PRICE.people_search)}`); if (num(agentTotalFinance) > 0) bits.push(`finance ${q(agentTotalFinance)}×${fmtRate(TOOL_PRICE.finance_search)}`); return `total = ${[...parts, ...bits].join(' + ')} = ${fmt(result.total)}`; } const parts = [`${q(num(agentInput))} × ${fmtRate(inR)}/1M`, `${q(agentOutput)} × ${fmtRate(outR)}/1M`]; const bits = []; if (num(agentSandboxSessions) > 0) bits.push(`sandbox ${qRun(agentSandboxSessions)}×${fmtRate(SANDBOX_SESSION)}`); if (num(agentSandboxSearches) > 0) bits.push(`sandbox-search ${qRun(agentSandboxSearches)}×${fmtRate(SANDBOX_SEARCH)}`); if (num(agentWebSearch) > 0) bits.push(`web ${qRun(agentWebSearch)}×${fmtRate(TOOL_PRICE.web_search)}`); if (num(agentFetchUrl) > 0) bits.push(`fetch ${qRun(agentFetchUrl)}×${fmtRate(TOOL_PRICE.fetch_url)}`); if (num(agentPeople) > 0) bits.push(`people ${qRun(agentPeople)}×${fmtRate(TOOL_PRICE.people_search)}`); if (num(agentFinance) > 0) bits.push(`finance ${qRun(agentFinance)}×${fmtRate(TOOL_PRICE.finance_search)}`); const inner = [...parts, ...bits].join(' + '); return `total = ( ${inner} ) × ${q(agentRunsPerMonth)} API calls = ${fmt(result.total * num(agentRunsPerMonth))}`; }; const sonarEstimateText = () => { const m = sonarModel; const isDR = sonarIsDeepResearch; if (usageMode === 'total') { const lines = ['Perplexity Sonar API cost estimate (total usage)', `Model: ${m.label}`, `Input tokens: ${qM(sonarTotalInputM)}M`, `Output tokens: ${qM(sonarTotalOutputM)}M`]; if (isDR) { lines.push(`Citation tokens: ${qM(sonarTotalCitationM)}M`, `Reasoning tokens: ${qM(sonarTotalReasoningM)}M`, `Search queries: ${q(sonarTotalSearches)}`); } else { lines.push(`Requests: ${q(sonarTotalRequests)} (search context ${sonarContext})`); } lines.push(`Estimated cost: ${fmt(result.total)}`, 'Estimate only. Final cost is metered from each response usage field.'); return lines.join('\n'); } const reqs = num(sonarRequests); const lines = ['Perplexity Sonar API cost estimate', `Model: ${m.label}`, `API calls: ${q(sonarRequests)}`, `Input tokens: ${q(sonarInput)}`, `Output tokens: ${q(sonarOutput)}`]; if (isDR) { lines.push(`Citation tokens: ${q(sonarCitation)}`, `Reasoning tokens: ${q(sonarReasoning)}`, `Search queries: ${q(sonarSearches)}`); } else { lines.push(`Search context: ${sonarContext} (request fee ${fmtRate(m.request[sonarContext])}/1K)`); } lines.push(`Per-call cost: ${fmt(result.total)}`, `Estimated cost: ${fmt(result.total * reqs)}`, 'Estimate only. Final cost is metered from each response usage field.'); return lines.join('\n'); }; const sonarFormulaText = () => { const m = sonarModel; const isDR = sonarIsDeepResearch; if (usageMode === 'total') { const parts = [`${qM(sonarTotalInputM)}M × ${fmtRate(m.input)}/1M`, `${qM(sonarTotalOutputM)}M × ${fmtRate(m.output)}/1M`]; if (isDR) parts.push(`citation ${qM(sonarTotalCitationM)}M × ${fmtRate(m.citation)}/1M`, `reasoning ${qM(sonarTotalReasoningM)}M × ${fmtRate(m.reasoning)}/1M`, `searches ${q(sonarTotalSearches)} × ${fmtRate(m.searchQueries)}/1K`); else parts.push(`request ${q(sonarTotalRequests)} × ${fmtRate(m.request[sonarContext])}/1K`); return `total = ${parts.join(' + ')} = ${fmt(result.total)}`; } const reqs = num(sonarRequests); const parts = [`${q(sonarInput)} × ${fmtRate(m.input)}/1M`, `${q(sonarOutput)} × ${fmtRate(m.output)}/1M`]; if (isDR) parts.push(`citation ${q(sonarCitation)} × ${fmtRate(m.citation)}/1M`, `reasoning ${q(sonarReasoning)} × ${fmtRate(m.reasoning)}/1M`, `searches ${q(sonarSearches)} × ${fmtRate(m.searchQueries)}/1K`); else parts.push(`request ${fmtRate(m.request[sonarContext])}/1K`); return `total = ( ${parts.join(' + ')} ) × ${q(sonarRequests)} API calls = ${fmt(result.total * reqs)}`; }; const estimateText = () => isSonar ? sonarEstimateText() : agentEstimateText(); const formulaText = () => isSonar ? sonarFormulaText() : agentFormulaText(); const renderRateChip = text =>

    {text}

    ; const sectionLabel = text =>

    {text}

    ; const POP_W = 256; const POP_H = 160; const renderInfo = (id, content) => { const open = openInfo === id; const btnId = `${id}-btn`; const place = open ? infoPlace : { x: 'left', y: 'down' }; const openWithPlacement = el => { let x = 'left', y = 'down'; if (el && typeof window !== 'undefined') { const r = el.getBoundingClientRect(); if (r.left + POP_W > window.innerWidth - 8) x = 'right'; if (r.bottom + POP_H > window.innerHeight - 8) y = 'up'; } setInfoPlace({ x, y }); setOpenInfo(id); }; const isFocusVisible = el => { try { return !!(el && el.matches && el.matches(':focus-visible')); } catch (e) { return false; } }; const withinInfo = node => { const el = infoRefs.current[id]; return !!(el && node && el.contains(node)); }; const xCls = place.x === 'right' ? 'right-0' : 'left-0'; const yCls = place.y === 'up' ? 'bottom-4 pb-2' : 'top-4 pt-2'; return { infoRefs.current[id] = el; }} onMouseEnter={() => openWithPlacement(document.getElementById(btnId))} onMouseLeave={() => { if (typeof document !== 'undefined' && withinInfo(document.activeElement)) return; setOpenInfo(cur => cur === id ? null : cur); }} onFocus={e => { if (e.target && e.target.id === btnId && isFocusVisible(e.target)) openWithPlacement(e.target); }} onBlur={e => { if (e.relatedTarget && withinInfo(e.relatedTarget)) return; setOpenInfo(cur => cur === id ? null : cur); }}> {open && {content} } ; }; const TOOL_INFO = { web_search: ['Performs a web search to retrieve current information', TOOL_PRICE.web_search], fetch_url: ['Fetches and extracts content from a specific URL', TOOL_PRICE.fetch_url], people_search: ['Looks up professionals, employees, and people', TOOL_PRICE.people_search], finance_search: ['Retrieves financial data and market information', TOOL_PRICE.finance_search] }; const toolInfo = key => renderInfo(`tool-${key}-info`, <>{TOOL_INFO[key][0]} · {fmtRate(TOOL_INFO[key][1])} per invocation.); const toolsSandboxPerCall = () => num(agentWebSearch) * TOOL_PRICE.web_search + num(agentFetchUrl) * TOOL_PRICE.fetch_url + num(agentPeople) * TOOL_PRICE.people_search + num(agentFinance) * TOOL_PRICE.finance_search + num(agentSandboxSessions) * SANDBOX_SESSION + num(agentSandboxSearches) * SANDBOX_SEARCH; const agentModelCost = m => { const {inRate, outRate} = rateFor(m); if (usageMode === 'total') return num(agentTotalInputM) * inRate + num(agentTotalOutputM) * outRate; return (tokenCost(num(agentInput), inRate) + tokenCost(num(agentOutput), outRate)) * num(agentRunsPerMonth); }; const renderModelTable = () => { const providers = [...new Set(AGENT_MODELS.map(x => x.group))]; const terms = modelFilter.toLowerCase().split(',').map(t => t.trim()).filter(Boolean); const rows = AGENT_MODELS.filter(m => (modelProvider === 'all' || m.group === modelProvider) && (terms.length === 0 || terms.some(t => m.id.toLowerCase().includes(t) || m.group.toLowerCase().includes(t)))).sort((a, b) => modelSort === 'name' ? a.id.localeCompare(b.id) : agentModelCost(a) - agentModelCost(b)); const selInRows = rows.some(r => r.id === agentModelId); const focusRow = ni => { const t = rows[ni]; if (!t) return; setAgentModelId(t.id); if (typeof document !== 'undefined') { const el = document.getElementById(`calc-model-opt-${ni}`); if (el) el.focus(); } }; const onRowKey = (e, i) => { let ni = null; if (e.key === 'ArrowDown') ni = Math.min(rows.length - 1, i + 1); else if (e.key === 'ArrowUp') ni = Math.max(0, i - 1); else if (e.key === 'Home') ni = 0; else if (e.key === 'End') ni = rows.length - 1; if (ni !== null) { e.preventDefault(); focusRow(ni); } }; return
    setModelFilter(e.target.value)} placeholder="Filter models… e.g. claude, gpt-5, gemini" aria-label="Filter models" className={`${fieldBase} w-full`} />
    {['all', ...providers].map(prov => { const active = modelProvider === prov; return ; })}
    Sort: {[['cost', 'Cost'], ['name', 'Name']].map(([key, txt]) => )}
    {rows.map((m, i) => { const sel = m.id === agentModelId; const {inRate, outRate} = rateFor(m); const isTabStop = sel || !selInRows && i === 0; return ; })} {rows.length === 0 && No models match your filter.}
    ; }; const renderModelRow = () => { const m = agentModel; const tiered = isTieredModel(m); const thresholdLabel = formatTokenThreshold(tierThresholdFor(m)); const rateRange = v => v && typeof v === 'object' ? `${fmtRate(v.low)} → ${fmtRate(v.high)}` : fmtRate(v); const cacheLabel = m.cache === null ? 'no cache' : m.cache === 'inputx0.1' ? 'cache 90% off input' : `cache ${fmtRate(m.cache)}`; const modelInfo = {tiered ? `Input ${rateRange(m.input)} · Output ${rateRange(m.output)} per 1M · ${cacheLabel} · tier flips above ${thresholdLabel} input tokens` : `Input ${fmtRate(m.input)} · Output ${fmtRate(m.output)} · ${cacheLabel} per 1M tokens`} {!isSingleProduct && Rates from the Agent API Models page } ; return
    {renderInfo('model-info', modelInfo)} {tiered && {usageMode === 'total' ? 'base rate' : num(agentInput) > tierThresholdFor(m) ? `>${thresholdLabel} tier` : `≤${thresholdLabel} tier`} } {renderModelTable()}
    ; }; const tokenMField = (label, value, setter, id) => renderNumberField(label, value, setter, id, { narrow: true, decimals: 3, slider: renderSliderBar(value, setter, id, { scale: 'log', min: 0, logMin: 0.1, max: 1000, labelId: `${id}-label`, valueText: v => `${qM(v)} million ${label.replace(' (M)', '').toLowerCase()}` }) }); const renderModeTabs = () => { if (!(isAgent || isSonar)) return null; const accent = PRODUCT_META[activeProduct].accent; const order = ['total', 'percall']; const labels = { total: 'Total', percall: 'Per call' }; const tabId = key => `calc-mode-${key}`; const focusMode = key => { setUsageMode(key); if (typeof document !== 'undefined') { const el = document.getElementById(tabId(key)); if (el) el.focus(); } }; const onKey = e => { const i = order.indexOf(usageMode); let next = null; if (e.key === 'ArrowRight') next = order[(i + 1) % order.length]; else if (e.key === 'ArrowLeft') next = order[(i - 1 + order.length) % order.length]; else if (e.key === 'Home') next = order[0]; else if (e.key === 'End') next = order[order.length - 1]; if (next !== null) { e.preventDefault(); focusMode(next); } }; return
    {order.map(key => { const selected = usageMode === key; return ; })}
    ; }; const totalModeInfo = id => renderInfo(id, <>Total mode: enter your aggregate usage directly, with no per-call multiplier. Tokens are in millions (M) — 1 = 1,000,000 tokens. Tiered models are estimated at their base rate here, since a per-request tier can’t be inferred from a total.); const renderAgentInputs = () => { const total = usageMode === 'total'; const toolRow = (label, value, setter, id, info) => renderNumberField(label, value, setter, id, { info, narrow: true, slider: renderSliderBar(value, setter, id, total ? { scale: 'log', min: 0, max: 1000000, labelId: `${id}-label`, valueText: v => `${q(v)} ${label.toLowerCase()} total` } : { scale: 'linear', min: 0, max: 20, step: 1, labelId: `${id}-label`, valueText: v => `${v} ${label.replace(' / call', '').toLowerCase()} per API call` }) }); const TT = total ? [[agentTotalSandboxSessions, setAgentTotalSandboxSessions], [agentTotalSandboxSearches, setAgentTotalSandboxSearches], [agentTotalWebSearch, setAgentTotalWebSearch], [agentTotalFetchUrl, setAgentTotalFetchUrl], [agentTotalPeople, setAgentTotalPeople], [agentTotalFinance, setAgentTotalFinance]] : [[agentSandboxSessions, setAgentSandboxSessions], [agentSandboxSearches, setAgentSandboxSearches], [agentWebSearch, setAgentWebSearch], [agentFetchUrl, setAgentFetchUrl], [agentPeople, setAgentPeople], [agentFinance, setAgentFinance]]; const [[sessV, setSessV], [srchV, setSrchV], [webV, setWebV], [fetchV, setFetchV], [peopleV, setPeopleV], [financeV, setFinanceV]] = TT; return
    {}
    {sectionLabel('Usage')} {total ? totalModeInfo('agent-total-info') : renderInfo('usage-info', <>Input = your prompt + context sent to the model on each API call; Output = the model’s response. Enter the average per API call. Output tokens usually cost more per 1M than input, though input is often several times larger than output. This estimate bills all input at the full rate; prompt caching (the per-model cache rate) lowers your real cost, so treat the total as an upper bound.)} {total ? <> {tokenMField('Input Tokens (M)', agentTotalInputM, setAgentTotalInputM, 'calc-agent-total-input')} {tokenMField('Output Tokens (M)', agentTotalOutputM, setAgentTotalOutputM, 'calc-agent-total-output')} : <> {renderNumberField('API Calls', agentRunsPerMonth, setAgentRunsPerMonth, 'calc-agent-runs', { narrow: true, min: 1, slider: renderSliderBar(agentRunsPerMonth, setAgentRunsPerMonth, 'calc-agent-runs', { scale: 'log', min: 1, max: 10000000, labelId: 'calc-agent-runs-label', valueText: v => `${q(v)} API calls` }) })} {renderNumberField('Input Tokens', agentInput, setAgentInput, 'calc-agent-input', { narrow: true, slider: renderSliderBar(agentInput, setAgentInput, 'calc-agent-input', { scale: 'log', min: 0, max: 1000000, labelId: 'calc-agent-input-label', valueText: v => `${q(v)} input tokens per API call` }) })} {renderNumberField('Output Tokens', agentOutput, setAgentOutput, 'calc-agent-output', { narrow: true, slider: renderSliderBar(agentOutput, setAgentOutput, 'calc-agent-output', { scale: 'log', min: 0, max: 1000000, labelId: 'calc-agent-output-label', valueText: v => `${q(v)} output tokens per API call` }) })} }
    {} {(() => { const toolCount = [sessV, srchV, webV, fetchV, peopleV, financeV].reduce((s, v) => s + num(v), 0); return
    {toolsOpen &&
    {toolRow('Sandbox sessions', sessV, setSessV, 'calc-agent-sandbox-sessions', renderInfo('sandbox-sessions-info', <>Isolated code-execution session · {fmtRate(SANDBOX_SESSION)} per session (≤20-min billing window). SDK searches run inside a session are billed separately below.))} {} {num(sessV) >= 1 &&
    {toolRow('Sandbox searches', srchV, setSrchV, 'calc-agent-sandbox-searches', renderInfo('sandbox-searches-info', <>SDK searches run inside a sandbox session · {fmtRate(SANDBOX_SEARCH)} per search.))}
    } {toolRow('Web search', webV, setWebV, 'calc-agent-web', toolInfo('web_search'))} {toolRow('Fetch URL', fetchV, setFetchV, 'calc-agent-fetch', toolInfo('fetch_url'))} {toolRow('People search', peopleV, setPeopleV, 'calc-agent-people', toolInfo('people_search'))} {toolRow('Finance search', financeV, setFinanceV, 'calc-agent-finance', toolInfo('finance_search'))}
    }
    ; })()} {renderModelRow()}
    ; }; const renderSearchInputs = () =>
    {renderNumberField('Requests', searchRequests, setSearchRequests, 'calc-search-requests', { narrow: true, min: 1, slider: renderSliderBar(searchRequests, setSearchRequests, 'calc-search-requests', { scale: 'log', min: 1, max: 10000000, labelId: 'calc-search-requests-label', valueText: v => `${q(v)} requests` }) })}

    The Search API charges per request only — no token costs. {fmtRate(SEARCH_PER_1K)} per 1,000 requests.

    ; const renderSonarContextControl = () => { const m = sonarModel; const sizes = [['low', 'Low'], ['medium', 'Medium'], ['high', 'High']]; return
    {renderInfo('sonar-context-info', <>How much web information is retrieved per query. Higher context = more comprehensive results and a higher per-request fee; token pricing is unchanged. Low is the default.)}
    {sizes.map(([key, txt]) => { const active = sonarContext === key; const fee = m.request ? m.request[key] : 0; return ; })}
    ; }; const renderSonarInputs = () => { const m = sonarModel; const isDR = sonarIsDeepResearch; const total = usageMode === 'total'; const rateChip = isDR ? `Input ${fmtRate(m.input)} · Output ${fmtRate(m.output)} · Citation ${fmtRate(m.citation)} · Reasoning ${fmtRate(m.reasoning)} /1M · Searches ${fmtRate(m.searchQueries)}/1K` : `Input ${fmtRate(m.input)} · Output ${fmtRate(m.output)} /1M · Request ${fmtRate(m.request[sonarContext])}/1K (${sonarContext})`; const tokenSlider = (value, setter, id, label) => renderSliderBar(value, setter, id, { scale: 'log', min: 0, max: 1000000, labelId: `${id}-label`, valueText: v => `${q(v)} ${label} per API call` }); return
    {renderInfo('sonar-model-info', <>Total cost per query = token costs + a per-request fee (Sonar, Sonar Pro, and Sonar Reasoning Pro). Sonar Deep Research instead bills citation and reasoning tokens plus per-search-query fees.)} {renderListbox('calc-sonar-model', sonarModelId, setSonarModelId, SONAR_MODELS, { renderOption: o => o.label })} {renderRateChip(rateChip)}
    {}
    {sectionLabel('Usage')} {total ? totalModeInfo('sonar-total-info') : renderInfo('sonar-usage-info', <>Input = your prompt + context sent on each API call; Output = the model’s response. Enter the average per API call. This estimate bills all tokens at the full rate, so treat the total as an upper bound.)} {total ? <> {tokenMField('Input Tokens (M)', sonarTotalInputM, setSonarTotalInputM, 'calc-sonar-total-input')} {tokenMField('Output Tokens (M)', sonarTotalOutputM, setSonarTotalOutputM, 'calc-sonar-total-output')} {isDR && <> {tokenMField('Citation Tokens (M)', sonarTotalCitationM, setSonarTotalCitationM, 'calc-sonar-total-citation')} {tokenMField('Reasoning Tokens (M)', sonarTotalReasoningM, setSonarTotalReasoningM, 'calc-sonar-total-reasoning')} {renderNumberField('Search Queries', sonarTotalSearches, setSonarTotalSearches, 'calc-sonar-total-searches', { narrow: true, slider: renderSliderBar(sonarTotalSearches, setSonarTotalSearches, 'calc-sonar-total-searches', { scale: 'log', min: 0, max: 10000000, labelId: 'calc-sonar-total-searches-label', valueText: v => `${q(v)} search queries total` }) })} } : <> {renderNumberField('API Calls', sonarRequests, setSonarRequests, 'calc-sonar-requests', { narrow: true, min: 1, slider: renderSliderBar(sonarRequests, setSonarRequests, 'calc-sonar-requests', { scale: 'log', min: 1, max: 10000000, labelId: 'calc-sonar-requests-label', valueText: v => `${q(v)} API calls` }) })} {renderNumberField('Input Tokens', sonarInput, setSonarInput, 'calc-sonar-input', { narrow: true, slider: tokenSlider(sonarInput, setSonarInput, 'calc-sonar-input', 'input tokens') })} {renderNumberField('Output Tokens', sonarOutput, setSonarOutput, 'calc-sonar-output', { narrow: true, slider: tokenSlider(sonarOutput, setSonarOutput, 'calc-sonar-output', 'output tokens') })} {isDR && <> {renderNumberField('Citation Tokens', sonarCitation, setSonarCitation, 'calc-sonar-citation', { narrow: true, slider: tokenSlider(sonarCitation, setSonarCitation, 'calc-sonar-citation', 'citation tokens') })} {renderNumberField('Reasoning Tokens', sonarReasoning, setSonarReasoning, 'calc-sonar-reasoning', { narrow: true, slider: tokenSlider(sonarReasoning, setSonarReasoning, 'calc-sonar-reasoning', 'reasoning tokens') })} {renderNumberField('Search Queries', sonarSearches, setSonarSearches, 'calc-sonar-searches', { narrow: true, slider: renderSliderBar(sonarSearches, setSonarSearches, 'calc-sonar-searches', { scale: 'linear', min: 0, max: 100, step: 1, labelId: 'calc-sonar-searches-label', valueText: v => `${q(v)} search queries per API call` }) })} } }
    {!isDR &&
    {total && renderNumberField('Requests', sonarTotalRequests, setSonarTotalRequests, 'calc-sonar-total-requests', { narrow: true, helper: 'Total requests — used only for the per-request fee.', slider: renderSliderBar(sonarTotalRequests, setSonarTotalRequests, 'calc-sonar-total-requests', { scale: 'log', min: 0, max: 10000000, labelId: 'calc-sonar-total-requests-label', valueText: v => `${q(v)} requests total` }) })} {renderSonarContextControl()}
    }
    ; }; const renderEmbeddingsInputs = () =>
    {sectionLabel('Usage')}
    {renderListbox('calc-emb-model', embModelId, setEmbModelId, EMB_MODELS, { renderOption: o => `${o.id} · ${o.dims} dims` })} {renderRateChip(`${fmtRate(embModel.rate)} per 1M tokens · ${embModel.dims} dims`)}
    {renderNumberField('Tokens per request', embTokens, setEmbTokens, 'calc-emb-tokens')}
    {renderNumberField('Requests', volume, setVolume, 'calc-emb-volume', { narrow: true, min: 1, slider: renderSliderBar(volume, setVolume, 'calc-emb-volume', { scale: 'log', min: 1, max: 10000000, labelId: 'calc-emb-volume-label', valueText: v => `${q(v)} requests` }) })}
    ; const renderInputs = () => { if (activeProduct === 'agent') return renderAgentInputs(); if (activeProduct === 'search') return renderSearchInputs(); if (activeProduct === 'sonar') return renderSonarInputs(); return renderEmbeddingsInputs(); }; const breakdownRow = (key, label, valueNode, mathNode, topBorder) =>
    {label} {valueNode}
    {mathNode && {mathNode}}
    ; const subRow = (key, label, math, value) =>
    {label} {value}
    {math && {math}}
    ; const renderBreakdownRows = () => { const rows = []; if (activeProduct === 'agent') { const r = result; if (isTotalMode) { const {inR, outR} = agentRates; if (r.inputCost > 0) rows.push(breakdownRow('t-in', 'Input tokens', fmt(r.inputCost), `${qM(agentTotalInputM)}M × ${fmtRate(inR)}/1M`)); if (r.outputCost > 0) rows.push(breakdownRow('t-out', 'Output tokens', fmt(r.outputCost), `${qM(agentTotalOutputM)}M × ${fmtRate(outR)}/1M`)); if (r.sandboxCost > 0) { rows.push(breakdownRow('t-sand', 'Sandbox', fmt(r.sandboxCost))); if (num(agentTotalSandboxSessions) > 0) rows.push(subRow('t-sess', 'Sandbox sessions', `${q(agentTotalSandboxSessions)} × ${fmtRate(SANDBOX_SESSION)}`, fmt(num(agentTotalSandboxSessions) * SANDBOX_SESSION))); if (num(agentTotalSandboxSearches) > 0) rows.push(subRow('t-ssrch', 'Sandbox searches', `${q(agentTotalSandboxSearches)} × ${fmtRate(SANDBOX_SEARCH)}`, fmt(num(agentTotalSandboxSearches) * SANDBOX_SEARCH))); } if (r.toolsCost > 0) { rows.push(breakdownRow('t-tools', 'Tool invocations', fmt(r.toolsCost))); if (r.tools.web > 0) rows.push(subRow('t-web', 'Web search', `${q(agentTotalWebSearch)} × ${fmtRate(TOOL_PRICE.web_search)}`, fmt(r.tools.web))); if (r.tools.fetch > 0) rows.push(subRow('t-fetch', 'Fetch URL', `${q(agentTotalFetchUrl)} × ${fmtRate(TOOL_PRICE.fetch_url)}`, fmt(r.tools.fetch))); if (r.tools.people > 0) rows.push(subRow('t-people', 'People search', `${q(agentTotalPeople)} × ${fmtRate(TOOL_PRICE.people_search)}`, fmt(r.tools.people))); if (r.tools.finance > 0) rows.push(subRow('t-finance', 'Finance search', `${q(agentTotalFinance)} × ${fmtRate(TOOL_PRICE.finance_search)}`, fmt(r.tools.finance))); } rows.push(breakdownRow('t-total', 'Total', fmt(r.total), undefined, true)); return rows; } const runs = num(agentRunsPerMonth); const {inR, outR} = agentRates; const thresholdLabel = formatTokenThreshold(r.tierThreshold); const tierSuffix = r.tiered ? r.tier === 'high' ? ` (>${thresholdLabel} tier)` : ` (≤${thresholdLabel} tier)` : ''; const modelPerRun = r.inputCost + r.outputCost; const toolsPerRun = r.toolsCost; const sandPerRun = r.sandboxCost; rows.push(breakdownRow('g-model', 'Model tokens', fmt(modelPerRun * runs), `${fmt(modelPerRun)}/call × ${q(runs)}`)); if (r.inputCost > 0) rows.push(subRow('in', `Input${tierSuffix}`, `${q(num(agentInput))} × ${fmtRate(inR)}/1M /call`, fmt(r.inputCost * runs))); if (r.outputCost > 0) rows.push(subRow('out', `Output${tierSuffix}`, `${q(agentOutput)} × ${fmtRate(outR)}/1M /call`, fmt(r.outputCost * runs))); if (sandPerRun > 0) { rows.push(breakdownRow('g-sand', 'Sandbox', fmt(sandPerRun * runs), `${fmt(sandPerRun)}/call × ${q(runs)}`)); if (num(agentSandboxSessions) > 0) rows.push(subRow('s-sess', 'Sandbox sessions', `${qRun(agentSandboxSessions)} × ${fmtRate(SANDBOX_SESSION)} /call`, fmt(num(agentSandboxSessions) * SANDBOX_SESSION * runs))); if (num(agentSandboxSearches) > 0) rows.push(subRow('s-srch', 'Sandbox searches', `${qRun(agentSandboxSearches)} × ${fmtRate(SANDBOX_SEARCH)} /call`, fmt(num(agentSandboxSearches) * SANDBOX_SEARCH * runs))); } if (toolsPerRun > 0) { rows.push(breakdownRow('g-tools', 'Tool invocations', fmt(toolsPerRun * runs), `${fmt(toolsPerRun)}/call × ${q(runs)}`)); if (r.tools.web > 0) rows.push(subRow('t-web', 'Web search', `${qRun(agentWebSearch)} × ${fmtRate(TOOL_PRICE.web_search)} /call`, fmt(r.tools.web * runs))); if (r.tools.fetch > 0) rows.push(subRow('t-fetch', 'Fetch URL', `${qRun(agentFetchUrl)} × ${fmtRate(TOOL_PRICE.fetch_url)} /call`, fmt(r.tools.fetch * runs))); if (r.tools.people > 0) rows.push(subRow('t-people', 'People search', `${qRun(agentPeople)} × ${fmtRate(TOOL_PRICE.people_search)} /call`, fmt(r.tools.people * runs))); if (r.tools.finance > 0) rows.push(subRow('t-finance', 'Finance search', `${qRun(agentFinance)} × ${fmtRate(TOOL_PRICE.finance_search)} /call`, fmt(r.tools.finance * runs))); } rows.push(breakdownRow('per-run', 'Per API call', fmt(r.total), undefined, true)); rows.push(breakdownRow('monthly', 'Total', fmt(r.total * runs), `${fmt(r.total)} × ${q(runs)}`)); return rows; } if (activeProduct === 'search') { rows.push(breakdownRow('req', 'Requests', fmt(result.total), `${q(searchRequests)} × ${fmtRate(SEARCH_PER_1K)} /1K`)); return rows; } if (activeProduct === 'sonar') { const r = result; const m = sonarModel; if (isTotalMode) { rows.push(breakdownRow('st-in', 'Input tokens', fmt(r.inputCost), `${qM(sonarTotalInputM)}M × ${fmtRate(m.input)}/1M`)); rows.push(breakdownRow('st-out', 'Output tokens', fmt(r.outputCost), `${qM(sonarTotalOutputM)}M × ${fmtRate(m.output)}/1M`)); if (r.deepResearch) { rows.push(breakdownRow('st-cit', 'Citation tokens', fmt(r.citationCost), `${qM(sonarTotalCitationM)}M × ${fmtRate(m.citation)}/1M`)); rows.push(breakdownRow('st-rea', 'Reasoning tokens', fmt(r.reasoningCost), `${qM(sonarTotalReasoningM)}M × ${fmtRate(m.reasoning)}/1M`)); rows.push(breakdownRow('st-srch', 'Search queries', fmt(r.searchesCost), `${q(sonarTotalSearches)} × ${fmtRate(m.searchQueries)}/1K`)); } else { rows.push(breakdownRow('st-req', `Request fee (${sonarContext})`, fmt(r.requestFee), `${q(sonarTotalRequests)} × ${fmtRate(m.request[sonarContext])}/1K`)); } rows.push(breakdownRow('st-total', 'Total', fmt(r.total), undefined, true)); return rows; } const reqs = num(sonarRequests); rows.push(breakdownRow('s-in', 'Input tokens', fmt(r.inputCost), `${q(sonarInput)} × ${fmtRate(m.input)}/1M /call`)); rows.push(breakdownRow('s-out', 'Output tokens', fmt(r.outputCost), `${q(sonarOutput)} × ${fmtRate(m.output)}/1M /call`)); if (r.deepResearch) { rows.push(breakdownRow('s-cit', 'Citation tokens', fmt(r.citationCost), `${q(sonarCitation)} × ${fmtRate(m.citation)}/1M /call`)); rows.push(breakdownRow('s-rea', 'Reasoning tokens', fmt(r.reasoningCost), `${q(sonarReasoning)} × ${fmtRate(m.reasoning)}/1M /call`)); rows.push(breakdownRow('s-srch', 'Search queries', fmt(r.searchesCost), `${q(sonarSearches)} × ${fmtRate(m.searchQueries)}/1K /call`)); } else { rows.push(breakdownRow('s-req', `Request fee (${sonarContext})`, fmt(r.requestFee), `${fmtRate(m.request[sonarContext])}/1K /call`)); } rows.push(breakdownRow('s-per', 'Per API call', fmt(r.total), undefined, true)); if (reqs !== 1) rows.push(breakdownRow('s-total', 'Total', fmt(r.total * reqs), `${fmt(r.total)} × ${q(reqs)}`)); return rows; } rows.push(breakdownRow('tok', 'Tokens', fmt(result.total), `${q(embTokens)} × ${fmtRate(embModel.rate)} /1M`)); const showMonthly = num(volume) !== 1; rows.push(breakdownRow('per-call', 'Per request', fmt(unitCost), undefined, true)); if (showMonthly) rows.push(breakdownRow('monthly', 'Total', fmt(projected), `${fmt(unitCost)} × ${q(volume)}`)); return rows; }; const renderBreakdown = () => { if (!showBreakdown) return null; return
    {renderBreakdownRows()}
    ; }; const renderResult = () => { const scaled = !isSearch && !isAgent && num(volume) > 1; const eyebrow = isSearch ? `Estimated cost for ${q(searchRequests)} requests` : 'Estimated cost'; const nowrap = s => {s}; let caption; if (isTotalMode) { const inM = isAgent ? agentTotalInputM : sonarTotalInputM; const outM = isAgent ? agentTotalOutputM : sonarTotalOutputM; caption = <>Total usage · {nowrap(`${qM(inM)}M in`)} · {nowrap(`${qM(outM)}M out`)}; } else if (isAgent) { const hasToolCost = (result.toolsCost || 0) + (result.sandboxCost || 0) > 0; caption = <>{fmt(result.total)} per API call {nowrap(hasToolCost ? '(model + tools)' : '(model tokens)')} × {nowrap(`${q(agentRunsPerMonth)} API calls`)}; } else if (isSearch) caption = <>{q(searchRequests)} requests · {nowrap(`${fmt(unitCost)} per request`)}; else if (isSonar) caption = <>{fmt(result.total)} per API call × {nowrap(`${q(sonarRequests)} API calls`)}; else caption = scaled ? <>{fmt(unitCost)} per request × {nowrap(`${q(volume)} requests`)} : 'per request'; const portalLink = API Portal; const finePrint = isSingleProduct ? <>Estimate only. Final cost is metered from each response’s usage field - view it in the {portalLink}. : <>Estimate only. Final cost is metered from each response’s usage field - view it in the {portalLink}. See the full rate tables below.; return
    {liveText}
    {}

    {eyebrow}

    {displayProjected}

    {caption}

    {isAgent &&

    Model: {agentModel.id.replace(/^[^/]+\//, '')}

    } {isSonar &&

    Model: {sonarModel.label}

    }
    {renderBreakdown()}
    {} {(isAgent || isSonar) &&
    {showFormula &&

    {formulaText()}

    }
    }

    {finePrint}

    ; }; const renderTabBar = () => { if (isSingleProduct) { const meta = PRODUCT_META[activeProduct]; return
    ; } const onTabKeyDown = e => { const idx = PRODUCTS.indexOf(activeProduct); let next = null; if (e.key === 'ArrowRight') next = (idx + 1) % PRODUCTS.length; else if (e.key === 'ArrowLeft') next = (idx - 1 + PRODUCTS.length) % PRODUCTS.length; else if (e.key === 'Home') next = 0; else if (e.key === 'End') next = PRODUCTS.length - 1; if (next !== null) { e.preventDefault(); const id = PRODUCTS[next]; setActiveProduct(id); if (typeof document !== 'undefined') { const el = document.getElementById(`calc-tab-${id}`); if (el) el.focus(); } } }; return
    {PRODUCTS.map(p => { const meta = PRODUCT_META[p]; const selected = activeProduct === p; return ; })}
    ; }; const renderPresetRow = () => { if (!isAgent || usageMode === 'total') return null; return
    Common tasks
    {AGENT_PRESETS.map(p => )}
    ; }; const panelProps = isSingleProduct ? {} : { role: 'tabpanel', id: `calc-panel-${activeProduct}`, 'aria-labelledby': `calc-tab-${activeProduct}` }; return
    {renderTabBar()}
    {renderModeTabs()} {renderPresetRow()}
    {renderInputs()}
    {renderResult()}
    ; }; This page shows **pricing information** to help you understand API costs. For **billing setup**, payment methods, and usage monitoring, visit the [Admin section](/docs/getting-started/api-groups). For **rate limits**, see the [Rate Limits & Usage Tiers](/docs/admin/rate-limits-usage-tiers) page. ## Estimate your cost ## Agent API Pricing The Agent API provides access to third-party models from providers including OpenAI, Anthropic, Google, xAI, Z.AI, Moonshot AI, and NVIDIA with **transparent, token-based pricing** at direct provider rates with no markup. ### Model Pricing Agent API pricing varies by provider and model, with each provider offering multiple models at different price points. See the full pricing breakdown for all available models from OpenAI, Anthropic, Google, xAI, Z.AI, Moonshot AI, and NVIDIA, including cache rates and provider documentation links on the [Agent API Models page](/docs/agent-api/models). ### Tool Pricing When using tools with the Agent API: | Tool | Price | Description | | -------------------- | :---------------------: | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **`web_search`** | \$0.005 per invocation | Performs web searches to retrieve current information | | **`fetch_url`** | \$0.0005 per invocation | Fetches and extracts content from specific URLs | | **`people_search`** | \$0.005 per invocation | Looks up professionals, employees, and people. \$5 per 1,000 tool invocations | | **`finance_search`** | \$0.005 per invocation | Retrieves financial data and market information. \$5 per 1,000 tool invocations | | **`sandbox`** | \$0.03 per session | Isolated container for executing code during an Agent API request. A session covers up to 20 minutes of active use for billing purposes — this is the billing window, not a runtime cap. SDK search queries made from inside the sandbox are billed at \$0.005 per request (same as `web_search`). | Most tool costs are per invocation. `sandbox` is billed per container session — a 20-minute billing window per container, not a runtime cap — plus per SDK search query made from inside it. Tool costs are separate from model token costs. ## Search API Pricing | API | Price per 1K requests | Description | | -------------- | :-------------------: | ---------------------------------------------- | | **Search API** | \$5.00 | Raw web search results with advanced filtering | **No token costs:** Search API charges per request only, with no additional token-based pricing. ## Sonar API Pricing **Total cost per query** = Token costs + Request fee (varies by search context size, applies to Sonar, Sonar Pro, and Sonar Reasoning Pro models only) ## Token Pricing **Token pricing** is based on the number of tokens in your request and response. | Model | Input Tokens (\$/1M) | Output Tokens (\$/1M) | Citation Tokens (\$/1M) | Search Queries (\$/1K) | Reasoning Tokens (\$/1M) | | ----------------------- | :------------------: | :-------------------: | :---------------------: | :--------------------: | :----------------------: | | **Sonar** | \$1 | \$1 | - | - | - | | **Sonar Pro** | \$3 | \$15 | - | - | - | | **Sonar Reasoning Pro** | \$2 | \$8 | - | - | - | | **Sonar Deep Research** | \$2 | \$8 | \$2 | \$5 | \$3 | ## Request Pricing by Search Context Size **Search context** determines how much web information is retrieved. Higher context = more comprehensive results. The following table shows the request fee for each model for every **1000 requests**. | Model | Low Context Size | Medium Context Size | High Context Size | | ----------------------- | :--------------: | :-----------------: | :---------------: | | **Sonar** | \$5 | \$8 | \$12 | | **Sonar Pro** | \$6 | \$10 | \$14 | | **Sonar Reasoning Pro** | \$6 | \$10 | \$14 | * **Low**: (default) fastest, cheapest * **Medium**: Balanced cost/quality * **High**: Maximum search depth, best for research [Learn more about search context →](https://docs.perplexity.ai/docs/sonar/filters#context-size-control) ## Pro Search Pricing (Pro Search for Sonar Pro) **Pro Search** enhances Sonar Pro with automated tool usage and multi-step reasoning. When enabled, the model can perform multiple web searches and fetch URL content to answer complex queries. [Learn more about Pro Search here](/docs/sonar/pro-search/quickstart). Pro Search requires `stream: true` and is enabled via the `search_type` parameter in `web_search_options`. ### Search Type Options | Search Type | Description | Request Fee (per 1K) | | ----------- | -------------------------------------------------- | :----------------------: | | **`fast`** | (default) Standard Sonar Pro behavior | \$6 / \$10 / \$14 | | **`pro`** | Multi-step tool usage for complex queries | \$14 / \$18 / \$22 | | **`auto`** | Automatic classification based on query complexity | Varies by classification | Request fees vary by search context size (Low / Medium / High). Token pricing remains the same as standard Sonar Pro (\$3 per 1M input, \$15 per 1M output). ## Embeddings API Pricing Generate high-quality text embeddings for semantic search, retrieval-augmented generation (RAG), and other machine learning applications. ### Standard Embeddings | Model | Dimensions | Price (\$/1M tokens) | | -------------------- | :--------: | :------------------: | | `pplx-embed-v1-0.6b` | 1024 | \$0.004 | | `pplx-embed-v1-4b` | 2560 | \$0.03 | ### Contextualized Embeddings | Model | Dimensions | Price (\$/1M tokens) | | ---------------------------- | :--------: | :------------------: | | `pplx-embed-context-v1-0.6b` | 1024 | \$0.008 | | `pplx-embed-context-v1-4b` | 2560 | \$0.05 | Learn how to use the Embeddings API for semantic search, RAG, and more. ### Input Tokens The number of tokens in your prompt or message to the API. This includes: * Your question or instruction * Any context or examples you provide * System messages and formatting **Example:** "What is the weather in New York?" = \~8 input tokens ### Output Tokens The number of tokens in the API's response. This includes: * The generated answer or content * Any explanations or additional context * Search results and references **Example:** "The weather in New York is currently sunny with a temperature of 72°F." = \~15 output tokens ### Citation Tokens Tokens used specifically for generating search results and references in responses. Only applies to **Sonar Deep Research** model. **Example:** Including source links, reference numbers, and bibliographic information ### Search Context Size vs Context Window **Search context size** is *not* the same as the **context window**. * **Search context size**: How much web information is retrieved during search (affects request pricing) * **Context window**: Maximum tokens the model can process in one request (affects token limits) ### Search Queries The number of individual searches conducted by **Sonar Deep Research** during query processing. This is separate from your initial user query. * The model automatically determines how many searches are needed * You cannot control the exact number of search queries * The `reasoning_effort` parameter influences the number of searches performed * Only applies to **Sonar Deep Research** model ### Reasoning Tokens Tokens used for step-by-step logical reasoning and problem-solving. Only applies to **Sonar Deep Research** model. **Example:** Breaking down a complex math problem into sequential steps with explanations **Token Calculation:** 1 token ≈ 4 characters in English text. The exact count may vary based on language and content complexity. ## Cost Examples **Sonar** • 500 input + 200 output tokens | Component | Cost | | ------------- | ------------ | | Input tokens | \$0.0005 | | Output tokens | \$0.0002 | | Request fee | \$0.005 | | **Total** | **\$0.0057** | | Component | Cost | | ------------- | ------------ | | Input tokens | \$0.0005 | | Output tokens | \$0.0002 | | Request fee | \$0.008 | | **Total** | **\$0.0087** | | Component | Cost | | ------------- | ------------ | | Input tokens | \$0.0005 | | Output tokens | \$0.0002 | | Request fee | \$0.012 | | **Total** | **\$0.0127** | **Sonar Deep Research** | Component | Cost | | ------------------------ | -------------- | | Input tokens (33) | \$0.000066 | | Output tokens (7163) | \$0.057304 | | Citation tokens (20016) | \$0.040032 | | Reasoning tokens (73997) | \$0.221991 | | Search queries (18) | \$0.09 | | **Total** | **\$0.409393** | | Component | Cost | | ------------------------- | ----------- | | Input tokens (7) | \$0.00 | | Output tokens (3847) | \$0.031 | | Citation tokens (47293) | \$0.095 | | Reasoning tokens (308156) | \$0.924 | | Search queries (28) | \$0.14 | | **Total** | **\$1.190** | | Component | Cost | | ------------------------- | ----------- | | Input tokens (8) | \$0.00 | | Output tokens (4435) | \$0.035 | | Citation tokens (58196) | \$0.116 | | Reasoning tokens (339594) | \$1.019 | | Search queries (30) | \$0.15 | | **Total** | **\$1.320** | ## Purchase Options Purchase API credits through AWS Marketplace with consolidated billing and enterprise procurement. Fill out our enterprise inquiry form to discuss custom pricing, dedicated support, and enterprise features for teams and organizations.