Arcee AI has introduced Trinity Large Thinking, an open source reasoning model with a massive 262,144 token context window. Released on April 1, 2026, the model shows strong performance in PinchBench benchmarks and various reasoning tasks. Available through OpenRouter, the platform intelligently routes requests to optimal providers while maintaining uptime through fallback mechanisms.

Trinity Large Thinking is priced at $0.25 per million input tokens and $0.90 per million output tokens, with reasoning tokens tracked separately to display the model's internal thinking process. OpenRouter normalizes requests across providers and supports continued reasoning by preserving complete reasoning_details when passing messages back to the model. The model is free in open claw for the first five days after launch.

The combination of large context window, competitive pricing, and reasoning capabilities makes Trinity Large Thinking suitable for complex agentic workloads. Developers can access the model through OpenRouter's platform, which provides standardized interfaces across different providers while allowing applications to appear on OpenRouter leaderboards by setting appropriate headers.