Different research teams have demonstrated jailbreaks against ChatGPT, DeepSeek, and Alibaba’s Qwen AI models.
The French firm's own Thursday release, "Mistral Small 3", is "competitive with larger models" including Meta's Llama or Qwen ...