Trinity-Large-Thinking by Arcee

Arcee just dropped Trinity-Large-Thinking, and they're making the wild claim that it matches Opus 4.6 performance at 96% less cost. That's the kind of promise that sounds amazing on paper until you actually test it, but if they pulled it off, that's a massive signal for every startup that's been hemorrhaging money on API calls. The "open model" part means developers can actually inspect and self-host it rather than being locked into another proprietary provider, which is the real move here if the quality holds up.