blockrun_chat
Query AI models from multiple providers with routing modes for speed, cost, or specialization. Supports multi-turn chat and model-specific options.
Instructions
Get a second opinion from another AI model, or use a specialized model for a specific task.
Notable modes:
mode:"glm" → Zhipu GLM-5 / GLM-5-Turbo ($0.001/call, excellent for coding tasks, pays via USDC on BlockRun)
mode:"coding" → GLM-5 first, then code-specialized models
mode:"cheap" → GLM-5, NVIDIA free, DeepSeek
mode:"reasoning" → Claude Opus, o3, o1, deepseek-reasoner
mode:"free" → NVIDIA models (no cost)
routing:"smart" → auto-select via ClawRouter
Pick directly: model:"zai/glm-5", model:"openai/o3", model:"nvidia/deepseek-v4-flash" (free).
Run blockrun_models to see all available models with pricing.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| mode | No | Routing mode: glm = Zhipu GLM-5/GLM-5-Turbo ($0.001/call, great for coding), coding = GLM-5 + code models, cheap = GLM-5 + budget, free = NVIDIA only (ignored if model specified) | |
| stop | No | Up to 4 stop sequences; generation halts when any is produced | |
| model | No | Specific model ID (e.g., 'zai/glm-5', 'openai/o3') | |
| system | No | Optional system prompt | |
| message | Yes | Your message to the AI | |
| routing | No | Set to "smart" to auto-select the optimal model via ClawRouter (14-dimension AI routing) | |
| agent_id | No | Agent identifier. If a budget was delegated for this agent_id via blockrun_wallet action:'delegate', spending is tracked and enforced. The agent is hard-stopped when its budget is exhausted. | |
| messages | No | Conversation history for multi-turn context. When provided, 'message' is appended as the final user turn. Use with explicit 'model' param (defaults to 'openai/gpt-5.5' if not specified). Note: if you include a role:'system' entry in messages[], do not also pass the system param to avoid duplicate system messages. | |
| thinking | No | Anthropic extended thinking. Only honored for anthropic/claude-* models — these go direct to the native /v1/messages endpoint and the response includes verbatim type:'thinking' blocks with their original signature. Ignored for non-Claude models (no native thinking channel). | |
| max_tokens | No | Max tokens in response | |
| temperature | No | Creativity 0-2 | |
| response_format | No | Set to 'json_object' to force valid JSON output (no markdown fences). Works across all providers. | |
| routing_profile | No | Cost/quality profile for ClawRouter: "eco" (budget), "auto" (balanced, default), "premium" (best quality). Note: "free" maps to "auto" (the SDK dropped the free profile) and still settles a PAID model — for zero-cost generation use mode:"free" or model:"nvidia/...". Only applies when routing:"smart". | auto |