Mistral 7B Instruct v3
Mistral 7B v3 is the permissively-licensed answer to the 8B class. If the Meta license is a problem for your product, this is the fallback with almost no quality sacrifice — and its multilingual quality (30+ languages) is a real edge.
Who it is for: multilingual chat and RAG pipelines where Apache 2.0 matters, plus any 8 GB–16 GB GPU setup.
Specifications
| Type | llm |
|---|---|
| Parameters | 7.3B |
| Context window | 128k |
| License | Apache 2.0 |
| VRAM recommended | 16 GB (Q8) GB |
| VRAM minimum | 8 GB (Q4) GB |
Pros
- Permissive Apache 2.0 license
- Strong multilingual support
- 128k context at 7B size
Cons
- Slightly behind Llama 8B on tool calling
- Fewer community fine-tunes than the Llama family