TokenPost·

Anthropic’s FlashProve Generates Llama 4 Proofs at About 110 Tokens Per Second

Your call?

The benchmark used one H100 GPU, with proofs measuring about 19 megabytes per 1,024-token request.

Share

More news