Arcee<\/a>, a tiny 26-person U.S. startup that built a massive, 400B-parameter open source LLM on a $20 million shoestring budget<\/a>, has released its new reasoning model. Arcee calls the model Trinity Large Thinking \u2014 and it\u2019s the most capable open-weight model \u201cever released by a non-Chinese company,\u201d claims CEO Mark McQuade to TechCrunch.<\/p>\n As that comment implies, Arcee has a goal that I can\u2019t help but root for: It wants to give U.S. and Western companies a model that gives them no reason to use a Chinese-based one.<\/p>\n While Chinese models are extremely capable, they are perceived as risky<\/a>, putting power, and perhaps data, into the hands of a government that doesn\u2019t share all of the Western world\u2019s ideals.<\/p>\n With Arcee, companies can download the model, train it to their own needs, and use it on premises. Companies can also use Arcee\u2019s cloud-hosted version, accessible via API.<\/p>\n While Arcee\u2019s models are not outperforming the closed source models from the big labs like Anthropic or OpenAI, they\u2019re not being held hostage by the whims of those giants, either. <\/p>\n For instance, Claude, with its exceptional abilities to code, has been a popular choice for users of open source AI agent tool OpenClaw. But Anthropic pulled the rug out from them last week when it told users that their Anthropic subscriptions will no longer cover OpenClaw usage<\/a> \u2014 they will have to pay additionally for that. (In February, OpenClaw creator Peter Steinberger said he was\u00a0joining Anthropic\u2019s biggest rival, OpenAI<\/a>.)<\/p>\n In contrast, McQuade proudly points to data from OpenRouter<\/a> that says it has become one of the top models used with OpenClaw.<\/p>\n Techcrunch event<\/p>\n \n\t\t\t\t\t\t\t\t\tSan Francisco, CA<\/span> So, how good is Trinity Large Thinking? It is comparable to some of the other top open source models, according to the benchmark results it shared with TechCrunch.<\/p>\n As we previously reported<\/a>, it is not a head-to-head threat to the big cheese among U.S.-built open models: Meta\u2019s Llama 4. But it also doesn\u2019t have the odd, not-really open source license issues<\/a> of Meta\u2019s model. All of Arcee\u2019s Trinity models are released under the gold standard for OS licenses, Apache 2.0.<\/p>\n Just to be clear, there are also countless other U.S. startups offering open source models and, as a fan of the ingenuity of startups, I\u2019m rooting for them, too.<\/p>\n<\/div>\n
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 13-15, 2026<\/span>\n\t\t\t\t\t\t\t<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n![\"Arcee](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2026\/04\/Arcee.-Trinity-large-thinking-benchmark1.jpeg?w=680\")