Skip to main content

Overview

finance_search lets the model pull structured financial and market data for public companies, ETFs, and related instruments. The model decides which fields to fetch based on your prompt. Use it when one answer needs more than one type of financial data, such as valuation, earnings, and context for the same company or list of companies.

Capabilities

Data areaWhat it includes
Company basicsQuotes, profiles, peers, and market metadata
FinancialsIncome statement, balance sheet, cash flow (quarterly and annual), key ratios
Valuation and pricingCurrent/near-real-time pricing, 1-minute to 1-month OHLCV ranges, pre-market and after-hours data
EarningsLast earnings call transcript, report filings, beat/miss history, guidance discussion
Segment and KPI trackingRevenue/profit by segment, geography, ARPU, subscriber counts, GMV, and other operating metrics
Analyst coverageForward revenue and EPS estimates, cover count, historical estimate changes
Market activityTop gainers, top losers, and most active symbols
Ownership and corporate actionsInsider activity, ticker-level metadata, splits, and related market events
ETF and index detailsTop constituents, shares, weights, and market values

Coverage

finance_search coverage depends on the symbol, exchange, asset class, geography, and source data availability.
Coverage areaCurrent guidance
Ticker and symbol coveragePublicly traded companies, ETFs, and related instruments when a supported symbol can be resolved.
Asset classesPublic equities and ETFs.
Market data may be delayed, incomplete, or unavailable for some symbols. finance_search is for informational use only and does not provide investment, legal, tax, or financial advice.

Quickstart

Add finance_search to the tools array.

Example Prompts

  • Full company brief: “Give me a complete NVIDIA snapshot: valuation, segment revenue for the latest quarter, and management’s latest commentary on margins guidance.”
  • Compare companies in one request: “Compare Apple, Microsoft, and Alphabet on revenue growth, operating margin, and forward P/E for the latest fiscal year.”
  • Earnings + reaction context: “Summarize Tesla’s last earnings call, include actual vs consensus, and describe how the stock and analyst targets moved after publication.”

Prompt Guidance

finance_search works best when the prompt states the outcome, not the data shape.
  • Start with the business question first, then include the company or ticker.
  • Add time windows when relevant (latest quarter, fiscal year to date, last 30 days).
  • Let the tool decide which specific report fields to retrieve.
Start with the configuration that matches the shape of the finance question.
ConfigurationBest forLatencyQualityCost
Live Market Data and QuotesReal-time prices, quotes, and latest figuresFastGoodLow
Single-Company Historical LookupsBasic historical financials for one company or tickerBalancedHighMedium
Multi-Step Financial ResearchCross-company comparisons and complex financial analysisThoroughHighestHigh

Live Market Data and Quotes

Use this for time-sensitive answers that depend on real-time prices, quotes, or the latest market figures. It is the cheapest and fastest option while maintaining strong quality for live data lookups.

Single-Company Historical Lookups

Use this for a single company’s historical figures or basic questions that benefit from both structured finance data and web context. GPT-5.5 is strong at simple web search and token-efficient for historical lookups.

Multi-Step Financial Research

Use this for cross-company comparisons, longer historical investigations, and analysis that needs several tool calls across financial statements, filings, transcripts, and web sources. Opus performs best on complex multi-step reasoning when paired with the full tool suite.

Parameters

ParameterTypeRequiredDescription
typestringYesMust be "finance_search".

Response Shape

When finance_search runs, the response can include finance_results output items before the final assistant message. Each finance_results item includes the requested finance categories, ticker symbols, structured content, and source URLs when available. The final usage object includes token counts, cost details, and tool_calls_details.finance_search.invocation when tool-call usage is reported.

Pricing

finance_search is billed at $5 per 1,000 invocations. Model token usage is billed separately according to Agent API token pricing.
Pricing follows the same pattern as other tool calls: pay for invocations plus model tokens. See Pricing.

Limits / Quotas

finance_search uses the same Agent API request flow as other tools. Limits depend on your account tier, request configuration, and the number of tool invocations needed to answer the prompt. See Rate Limits & Usage Tiers for tier-based request limits.
Limit areaWhat to expect
API rate limitsAgent API rate limits apply to requests that use finance_search.
Tool invocationsEach time the model calls finance_search, it counts as one billable tool invocation. Multi-company or multi-step prompts may require more than one invocation.
Step limitsUse max_steps to cap multi-step requests that combine finance_search with tools such as web_search and fetch_url.
Output limitsLarge comparisons, long transcripts, and multi-year financial tables can hit max_output_tokens or response truncation settings.
Coverage limitsSome symbols, exchanges, regions, asset classes, or fields may be unavailable depending on source coverage and data freshness.

Next Steps

Web Search

Search the web for source-grounded context.

Fetch URL Content

Fetch full content from known URLs.

People Search

Search for professionals and employees.

Agent API Quickstart

Get started with the Agent API.