JarvisClaw

Chat completions (OpenAI-compatible). Supports streaming, function calling, vision.

Image generation.

Video generation.

Text-to-speech and audio generation.

Anthropic-native Messages API. Drop-in for Claude SDKs, Claude Code, and Anthropic-compatible clients.

Legacy OpenAI text completions for prompt-based clients.

Edit or rewrite text given an instruction.

Edit an existing image with a prompt and optional mask (inpainting).

WebSocket upgrade for real-time bidirectional streaming. Supports voice-to-voice and low-latency text with GPT-4o-realtime and Gemini Live models.

Lightweight Responses API variant — returns only the final assistant message without intermediate tool results. Lower cost for simple Q&A agents.

AI-powered prompt engineering: analyzes your prompt and returns an improved version with higher specificity and better structure.

Translate spoken audio in any language into English text.

Classify text for policy violations (content safety screening).

Text embeddings (OpenAI-compatible). Vectorize text for RAG, semantic search, and retrieval agents.

Rerank a set of documents against a query. High-frequency in RAG pipelines to sharpen retrieval relevance.

Speech-to-text transcription in the source language. For voice agents and audio ingestion pipelines.

Audio generation (music, sound effects, and generative audio).

Full Responses API. Stateful multi-turn agent runs with tool calls, reasoning, and intermediate results. The richest surface for autonomous agents.

Web/knowledge search proxy. Returns ranked results agents can ground answers on.

AIP — Agent Intent Protocol. Resolve a structured intent to the best-matching provider and route with confidence scores. Lets agents delegate 'what to call' instead of hard-coding endpoints.

AIP — resolve a free-text natural-language intent into ranked provider candidates. Ideal for LLM agents that emit plain language.

AIP — discover providers and capabilities across the federation for a given capability class. Returns the live catalog an agent can act on.

AIP — resolve intent and execute against the best provider in one call. End-to-end agent task completion.

AIP — resolve and execute an intent under a hard budget ceiling. Guarantees spend never exceeds the cap.

AIP — create a standing subscription that re-resolves and fulfills an intent on a schedule or trigger. Recurring agent tasks in one call.

A2A — Agent-to-Agent protocol entry point (JSON-RPC). Compatible clients (Claude, Cursor, A2A-aware agents) auto-discover this via /.well-known/agent.json and delegate tasks. Native agent interoperability surface.

Prediction market search with real-time odds. x402 double-hop settlement to upstream data providers.

creation — marketplace service, settled per call via x402.

voice — marketplace service, settled per call via x402.

rpc — marketplace service, settled per call via x402.

portrait — marketplace service, settled per call via x402.

compute — marketplace service, settled per call via x402.

defi — marketplace service, settled per call via x402.

surf — marketplace service, settled per call via x402.

dex — marketplace service, settled per call via x402.

realface — marketplace service, settled per call via x402.

data — marketplace service, settled per call via x402.

markets — marketplace service, settled per call via x402.

music — marketplace service, settled per call via x402.

phone — marketplace service, settled per call via x402.

prediction — marketplace service, settled per call via x402.

audio — marketplace service, settled per call via x402.

video — marketplace service, settled per call via x402.

exa — marketplace service, settled per call via x402.

Federated API catalogue — thousands of individually priced endpoints (search, blockchain data, OCR, code execution, image and document processing) callable by resource id and settled per call via x402. Browse the full list at listEndpoint; each row carries its own resource_id and price.

1031 Match

Agent Expert

Agent Fast

Agent Score

Agent Score