Mistral-Large-3-675B-Instruct-2512 released
Mistral released the full-precision instruct version of its 675B parameter Large 3 model on Hugging Face under Apache 2.0. The model is fine-tuned from Mistral-Large-3-675B-Base-2512, uses compressed-tensors, supports 12 languages, and the repository includes evaluation results.
PUBLISHED2025-11-28
OBSERVED2026-08-11
AGE9mo
SOURCES1
- 675B parameter instruct model, fine-tuned from Mistral-Large-3-675B-Base-2512
- Apache 2.0, compressed-tensors format, vLLM library
- 12 languages: en, fr, es, de, it, pt, nl, zh, ja, ko, ar
- Repository includes evaluation results (eval-results tag)
COMMUNITY
No curated reactions recorded for this event. Facts and takes are kept in separate layers — community context is added by hand, never blended into the record above.