> ## Documentation Index > Fetch the complete documentation index at: https://docs.dualmindlab.tech/llms.txt > Use this file to discover all available pages before exploring further. > ## Agent Instructions > DualMind Arena API base URL: https://api.dualmindlab.tech. Public endpoints only; admin routes are not part of this API. > Authenticate with the access token from an Arena session (a Supabase JWT) as `Authorization: Bearer `. There are no API keys yet; never invent one. > For battles call POST /api/arena/dualchat, then vote with POST /api/arena/model-vote using the returned comparisonId. Keep model names hidden from voters until they vote. > GET /api/arena/model-stats (leaderboard) and GET /health need no token. # Start a battle > Send one prompt to two models in parallel and get both answers plus a `comparisonId` to vote with. Leave `model1`/`model2` empty for a blind Battle (`random` or `topper`), or set both for Side by Side. **Keep it blind.** The response includes `agent1.model`, `agent2.model` and `arena.models` so you can reveal names after a vote. Keep them hidden from voters until they vote, as the Arena does. See [Run a battle](/api-reference/guides/run-a-battle). | You want | Send | | :- | :- | | A blind random battle | `prompt` only (or `selectionMode: "random"`) | | The leader vs a challenger | `selectionMode: "topper"` | | Two models you choose (Side by Side) | `model1` **and** `model2` | | To save it in a thread | `threadId` of a thread you own | Both models run in parallel with a 60 second limit each. A model that fails or times out is replaced by a fallback model, and its `message` starts with `[System Note]`. Use a client timeout of at least 120 seconds. ## OpenAPI ````yaml POST /api/arena/dualchat openapi: 3.1.0 info: title: DualMind Arena API version: 1.0.0 description: >- The REST API behind DualMind Arena: blind battles, votes, the leaderboard, models and threads. contact: name: DualMind Labs email: support@dualmindlab.tech url: https://dualmindlab.tech servers: - url: https://api.dualmindlab.tech description: Production security: - bearerAuth: [] tags: - name: Arena description: Battles and chat. - name: Voting description: Votes and the leaderboard. - name: Models description: Available models. - name: Threads description: Saved conversations and sharing. - name: Utilities description: Health and feature flags. paths: /api/arena/dualchat: post: tags: - Arena summary: Start a battle description: >- Send one prompt to two models in parallel and get both answers plus a `comparisonId` to vote with. Leave `model1`/`model2` empty for a blind Battle (`random` or `topper`), or set both for Side by Side. operationId: startBattle requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/BattleRequest' example: prompt: Write a haiku about distributed systems selectionMode: random responses: '200': description: Both responses and the comparison ID. content: application/json: schema: $ref: '#/components/schemas/BattleResponse' example: success: true agent1: object: ai.response output: type: message content: - type: output_text text: >- Packets cross the sea / two clocks disagree on now / quorum settles it success: true message: >- Packets cross the sea / two clocks disagree on now / quorum settles it model: name: llama-3.3-70b-versatile displayName: Llama 3.3 70B Versatile provider: groq prompt: Write a haiku about distributed systems selectionMode: random responseTimeMs: 1880 usage: promptTokens: 9 completionTokens: 19 totalTokens: 28 timestamp: '2026-10-17T12:00:00Z' agent2: object: ai.response output: type: message content: - type: output_text text: >- Nodes drift apart / consensus is a rumour / the log remembers success: true message: >- Nodes drift apart / consensus is a rumour / the log remembers model: name: gemini-2.5-flash displayName: Gemini 2.5 Flash provider: google prompt: Write a haiku about distributed systems selectionMode: random responseTimeMs: 1880 usage: promptTokens: 9 completionTokens: 17 totalTokens: 26 timestamp: '2026-10-17T12:00:00Z' comparisonId: 3f6c1a2b-9d4e-4f7a-8b1c-0e2d3f4a5b6c arena: comparison: winnerByLength: agent1 winnerByTokens: agent1 verdict: Agent 1 produced the longer, more token-heavy answer. agent1MessageLength: 70 agent2MessageLength: 61 agent1Tokens: 28 agent2Tokens: 26 models: agent1: llama-3.3-70b-versatile agent2: gemini-2.5-flash timestamp: '2026-10-17T12:00:02Z' totalResponseTimeMs: 1880 '400': description: Missing prompt, or only one of `model1`/`model2` set. content: application/json: schema: $ref: '#/components/schemas/AiError' example: object: ai.error code: INVALID_REQUEST message: Prompt is required and cannot be empty timestamp: '2026-10-17T12:00:00Z' '401': $ref: '#/components/responses/Unauthorized' '500': description: Unexpected server error. content: application/json: schema: $ref: '#/components/schemas/AiError' security: - bearerAuth: [] components: schemas: BattleRequest: type: object required: - prompt properties: prompt: type: string description: The prompt both models answer. Must not be empty. example: Write a haiku about distributed systems selectionMode: type: string enum: - random - topper default: random description: >- How the two models are picked when `model1` and `model2` are not set. `random`: two different random models. `topper`: the model with the best win rate against a random challenger. model1: type: string description: >- Left model (`modelName` or `modelId`). Set both `model1` and `model2` for a Side by Side comparison; setting only one is an error. model2: type: string description: Right model (`modelName` or `modelId`). system: type: string description: Optional system prompt sent to both models. maxTokens: type: integer minimum: 1 default: 4096 description: Maximum tokens per response. temperature: type: number minimum: 0 maximum: 2 default: 0.7 description: Sampling temperature for both models. threadId: type: string format: uuid description: Save this battle to a thread you own. BattleResponse: type: object properties: success: type: boolean comparisonId: type: string format: uuid description: Pass this to `POST /api/arena/model-vote`. agent1: $ref: '#/components/schemas/ChatResponse' description: Left response (vote `left`). agent2: $ref: '#/components/schemas/ChatResponse' description: Right response (vote `right`). arena: type: object description: >- Simple length and token heuristics plus the model names. Clients that want a blind experience should keep model names hidden until the user votes, as the Arena does. properties: comparison: type: object properties: winnerByLength: type: string enum: - agent1 - agent2 - tie winnerByTokens: type: string enum: - agent1 - agent2 - tie verdict: type: string userWinner: type: - string - 'null' description: Always `null` here; votes are submitted separately. agent1MessageLength: type: integer agent2MessageLength: type: integer agent1Tokens: type: integer agent2Tokens: type: integer models: type: object properties: agent1: type: string agent2: type: string timestamp: type: string format: date-time totalResponseTimeMs: type: integer AiError: type: object description: >- Error envelope used by the chat endpoints (and as an SSE event on the stream). properties: object: type: string const: ai.error code: type: string enum: - INVALID_REQUEST - API_ERROR - STREAM_ERROR message: type: string timestamp: type: string format: date-time ChatResponse: type: object properties: object: type: string const: ai.response success: type: boolean message: type: string description: The model's reply as plain text (Markdown). output: type: object description: The same reply in a structured content envelope. properties: type: type: string const: message content: type: array items: type: object properties: type: type: string const: output_text text: type: string model: $ref: '#/components/schemas/ModelInfo' prompt: type: string selectionMode: type: string enum: - automatic - manual - random - topper description: >- `automatic`/`manual` for single chat; `random`/`topper`/`manual` for battles. responseTimeMs: type: integer description: >- Wall-clock time of the request in milliseconds. In a battle both models run in parallel, so this is the time until both finished. usage: $ref: '#/components/schemas/Usage' timestamp: type: string format: date-time ModelInfo: type: object properties: name: type: string description: >- Provider model name, e.g. `llama-3.3-70b-versatile` or `@cf/meta/llama-3.1-8b-instruct`. displayName: type: string description: Human-friendly name. provider: type: string enum: - groq - google - cloudflare description: Inference provider. Usage: type: object properties: promptTokens: type: integer completionTokens: type: integer totalTokens: type: integer responses: Unauthorized: description: >- Missing, invalid or expired token. The body is empty; the `WWW-Authenticate` header explains why. securitySchemes: bearerAuth: type: http scheme: bearer bearerFormat: JWT description: >- The access token from your DualMind Arena session (a Supabase JWT). See Authentication. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.