Arcee Releases 400B‑Parameter Open‑Source AI Model
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Arcee built a 400‑billion‑parameter open‑source LLM on a $20 million budget and released it as the Trinity Large Thinking reasoning model.
- Mark McQuade claims Trinity Large Thinking is the most capable open‑weight model ever released by a non‑Chinese company.
- Arcee offers the model for on‑premises download and a cloud‑hosted API, letting companies run it without relying on big‑lab providers.
- OpenRouter data shows Trinity Large Thinking is among the top models used with the open‑source AI agent tool OpenClaw, overtaking Anthropic’s Claude after Anthropic stopped covering OpenClaw usage.
- Anthropic withdrew OpenClaw coverage from its Claude model, forcing users to pay extra, while OpenClaw creator Peter Steinberger joined OpenAI.
- Trinity Large Thinking benchmarks are comparable to other leading open‑source models, though it does not outpace closed‑source models from Anthropic or OpenAI.
- Arcee releases all Trinity models under the Apache 2.0 license, avoiding the licensing complications seen with Meta’s Llama 4.
Why it matters: Western firms now have a high‑capacity, on‑premises LLM that sidesteps reliance on Chinese models and the subscription constraints of big‑lab providers, while the open‑source licensing (Apache 2.0) avoids the legal ambiguities of alternatives like Meta’s Llama 4. This shifts some demand away from Anthropic and OpenAI’s closed‑source offerings.
