> ## Documentation Index > Fetch the complete documentation index at: https://lava.so/docs/llms.txt > Use this file to discover all available pages before exploring further. # Aggregated search across all 9 providers > Fan out a single query across 9 search providers in parallel — Tavily, Exa, Jina, Parallel, Perplexity Sonar, Firecrawl, Lava News, Serper, Brave Search — and return a unified normalized result set with explicit per-provider failure surfacing. Charges sum of upstream provider costs plus a $0.005 lava-search markup. Use this when you want the raw evidence pile to reason over yourself; use /search/answer instead when you want an LLM-synthesized answer with citations. ## OpenAPI ````yaml /openapi.json post /search/sources openapi: 3.0.0 info: title: Lava API description: >- Lava API enables businesses to implement usage-based billing for AI services. This API allows tracking, managing, and billing for third-party AI API usage through a forwarding system. version: 1.0.0 contact: name: Lava support url: https://www.lava.so/contact servers: - url: https://api.lava.so/v1/ description: Lava API v1 security: [] tags: - name: Authentication - name: Core Endpoints - name: Models - name: Services - name: Requests - name: Wallet - name: Payment Methods - name: Spend Keys - name: Usage - name: Checkout Sessions - name: Customers - name: Plans - name: Subscriptions - name: Meters - name: Webhooks - name: Secret Keys - name: Credit Bundles - name: Gateway Settings - name: Skills - name: News Data - name: Search - name: Enrich paths: /search/sources: post: tags: - Search summary: Aggregated search across all 9 providers description: >- Fan out a single query across 9 search providers in parallel — Tavily, Exa, Jina, Parallel, Perplexity Sonar, Firecrawl, Lava News, Serper, Brave Search — and return a unified normalized result set with explicit per-provider failure surfacing. Charges sum of upstream provider costs plus a $0.005 lava-search markup. Use this when you want the raw evidence pile to reason over yourself; use /search/answer instead when you want an LLM-synthesized answer with citations. requestBody: required: true content: application/json: schema: type: object required: - q properties: q: type: string description: The search query providers: type: array items: type: string nullable: true description: >- Optional subset of providers to query. Defaults to all 9. Valid ids: tavily, exa, jina, parallel, perplexity, firecrawl, lava_news, serper, brave_search. max_results_per_source: type: integer minimum: 1 maximum: 50 nullable: true description: Max results per provider (default 10, max 50) since: type: string format: date-time nullable: true description: >- ISO 8601 timestamp; only results published after this date are returned (provider-dependent — not all upstream APIs honor this). examples: medicineQuery: value: q: GLP-1 cardiovascular outcomes meta-analyses 2026 subsetQuery: value: q: post-quantum NIST standardization status providers: - exa - tavily - lava_news max_results_per_source: 5 responses: '200': description: >- Aggregated results across the requested providers, with explicit partial-failure surfacing. content: application/json: schema: type: object required: - results - sources_failed - cost properties: results: type: array items: type: object required: - source - title - url - snippet properties: source: type: string description: >- The provider id that returned this result (e.g. "tavily", "lava_news"). title: type: string url: type: string snippet: type: string published_at: type: string format: date-time nullable: true score: type: number nullable: true raw: type: object nullable: true description: >- Original upstream payload (for callers that want fields beyond the unified shape). sources_failed: type: array items: type: object required: - provider - reason properties: provider: type: string reason: type: string description: >- One of: timeout, upstream_error, parse_error, unknown_error. cost: type: object required: - breakdown - markup - total properties: breakdown: type: object additionalProperties: type: string description: >- Per-provider cost (decimal string) for every provider that returned results. markup: type: string description: Flat lava-search markup (always "0.05"). total: type: string description: sum(breakdown) + markup, as a decimal string. '400': description: >- Unknown provider id in request body. Response includes `error.unknown[]` listing the unrecognized ids. content: application/json: schema: $ref: '#/components/schemas/Error' '401': description: Unauthorized content: application/json: schema: $ref: '#/components/schemas/Error' '502': description: >- Every requested provider failed. Response includes `sources_failed[]` for diagnosis. No markup charged. content: application/json: schema: $ref: '#/components/schemas/Error' security: - Auth: [] components: schemas: Error: type: object properties: error: type: object properties: message: type: string description: Human-readable error message describing what went wrong example: Invalid authentication credentials code: type: string description: Machine-readable error code describing what went wrong example: forward_token_json_invalid status: type: integer description: HTTP status code of the error example: 400 issues: type: array items: type: object properties: path: type: array items: type: string description: Path to the field that caused the validation error message: type: string description: Error message describing the validation issue required: - path - message description: >- Optional array of specific validation issues with their paths and messages example: - path: - model message: 'Missing required field: model' - path: - messages - '0' - content message: Invalid message content format required: - message - code - status required: - error securitySchemes: Auth: type: http scheme: bearer description: >- Bearer token authentication used for standard API calls. Format: 'Bearer YOUR_API_KEY' ````