Anthropic and xAI Model Parameter Counts

Elon Musk says that Anthropic’s Claude Sonnet model has 1 trillion parameters and Claude Opus has 5 trillion parameters. XAI’s Grok 4.20, which has 0.5 trillion parameters and there are seven larger models being trained at XAI.

Grok Imagine V2
2 variants of 1-trillion-parameter models
2 variants of 1.5-trillion-parameter models
1 variant of a 6-trillion-parameter model
1 variant of a 10-trillion-parameter model

Claude Mythos is probably Mythos 10 trillion parameters.

1 thought on “Anthropic and xAI Model Parameter Counts”

  1. I dont know how they did it but Anthropic somewhat nailed it and others cant compete. First they had the best coding model. Then opus 4.5 on par with the best. Opus 4.6 the best with Gemini 3.1 pro and not just coding also general AI got so good. Now Mythos seems way above others. Grok is not bad, real time searching and answers are good, some things very good but with all that processing and training power they are not on pair with Gemini or Anthropic. These 2 have the lead and Musk is not winning. The ones I really dont want to win AI race – Facebook – Meta got decent model out, but they poached so many talented people from others, since they offered huge bonuses.

Comments are closed.