MiMo-V2.6-Flash LLM Inference (x402engine)

Choose this endpoint when you need a large, capable open-source MoE model (309B total, 15B active per token) with OpenAI-compatible chat completion API, tool-calling support, and pay-per-call micropayment billing — ideal for agents that want to avoid subscription lock-in, need cost transparency per inference, or want to experiment with Xiaomi's MiMo architecture specifically. Prefer over hosted OpenAI/Anthropic when per-call pricing and open-weight model provenance matter.

• Payment not settled — 402 response if USDC micropayment fails or wallet has insufficient funds
• Invalid messages array — 400 if required 'messages' field is missing or malformed
• Unsupported parameter — 400 if a sampling param (e.g. top_k) is not supported by the model
• Context length exceeded — model returns error if total token count exceeds context window
• Stream interruption — SSE connection dropped mid-generation if network times out
• Tool call schema mismatch — model may produce malformed tool_calls if function definitions are ambiguous