llm-meta-ai 0.2
Simon Willison updated `llm-meta-ai` for Meta’s new Muse Wonder models.
Meta AI released Muse Wonder and Muse Wonder Full, which Willison calls their new “most capable models.” He says they may be Meta’s first Fable class models, citing Artificial Analysis scoring Muse Wonder Max as equivalent to GPT-6.1 Sol. His test case is pelican images across six reasoning levels, with comparisons to Muse 1.3 and Muse 1.2. He notes Meta AI is shipping about one new model a month, with visible gains in his pelican benchmark. Simon Willison's note
Meta AI released Muse Wonder and Muse Wonder Full, which Willison calls their new “most capable models.” He says they may be Meta’s first Fable class models, citing Artificial Analysis scoring Muse Wonder Max as equivalent to GPT-6.1 Sol. His test case is pelican images across six reasoning levels, with comparisons to Muse 1.3 and Muse 1.2. He notes Meta AI is shipping about one new model a month, with visible gains in his pelican benchmark. Simon Willison's note
score 9