Mistral NeMo
Open Weights
Mistral AI
DISCONTINUED. Mistral and NVIDIA's jointly released 12B open model (July 2024), a drop-in upgrade to Mistral 7B with a 128K context window. Trained on multilingual and code data with a new Tekken tokenizer, it ships under Apache 2.0 in base and instruct versions plus an FP8 variant. Mistral deprecated the hosted model on May 22, 2026 with retirement on July 31, 2026, a date now passed; the Apache-2.0 weights remain on Hugging Face.
Strengths
- Open-weight 12B under Apache 2.0
- 128K context window
- Drop-in upgrade to Mistral 7B
- Strong multilingual and code data
- Runs on consumer GPUs (FP8 variant available)
Caveats
- DISCONTINUED as a hosted API - Mistral retired it on July 31, 2026
- Apache-2.0 weights remain available to self-host; Mistral names Ministral 3 8B as the replacement
- 12B ceiling below larger models
- Text-only (no vision input)
- Superseded by newer Mistral small models on some tasks
- Vendor
- Mistral AI
- Context Window
- 131,072 tokens
- License
- Apache-2.0
- Release Date
- 2024-07-18
- Modalities
- text
Capabilities
Vision
Audio
Video
Tool Use
Resources
Pricing
Open weights under Apache 2.0 - free to self-host. The hosted API model open-mistral-nemo-2407 was deprecated 5/22/2026 and retired 7/31/2026 (replacement: Ministral 3 8B).