Agentic Full-Task Runner (Plan → Generate → Verify)

Choose this endpoint when you need a fully self-contained agentic loop — not just generation, but planning and verification included — and can tolerate higher latency in exchange for higher-quality, self-checked outputs. Prefer it over single-shot generation endpoints when task correctness matters and you want the model to validate its own output before returning. Best for complex coding, reasoning, or writing tasks where one-pass generation is insufficient.

• Timeout on very complex tasks due to long-running nature of plan+generate+verify pipeline
• Malformed input schema may result in 400 bad request
• Cloudflare tunnel instability can cause connection resets or 502 errors on long-running requests
• Model may hallucinate verification success for incorrect outputs
• Rate limiting or payment failure (x402) blocks the call before execution begins
• Incomplete choices array in response if generation is truncated