Mistral Medium 3.5: 128B Open-Weight Frontier Coder
May 1, 2026
Mistral Medium 3.5 ships as a 128B dense open-weight model with a 256K context, 77.6% on SWE-Bench, and self-hosts on as few as four GPUs at $1.50/M tokens.
Mistral Medium 3.5 ships as a 128B dense open-weight model with a 256K context, 77.6% on SWE-Bench, and self-hosts on as few as four GPUs at $1.50/M tokens.
Mistral's Voxtral TTS is a 4B open-weight text-to-speech model with a 68.4% win rate vs ElevenLabs Flash v2.5. 9 languages, 3s voice cloning, $0.016/1K chars.
One email per week — courses, deep dives, tools, and AI experiments.
No spam. Unsubscribe anytime.