Mistral launches Small 3, a latency-optimized 24B-parameter model that it says is competitive with larger models such as Llama 3.3 70B or Qwen 32B (Mistral AI)

Mistral AI:
Mistral launches Small 3, a latency-optimized 24B-parameter model that it says is competitive with larger models such as Llama 3.3 70B or Qwen 32B  —  Apache 2.0, 81% MMLU, 150 tokens/s  —  Today we’re introducing Mistral Small 3, a latency-optimized 24B-parameter model released under the Apache 2.0 license.