Mistral 7B v0.3
Mistral 7B v0.3 extends the v0.2 vocabulary to 32,768 entries. The Instruct variant adds the v3 tokenizer and function calling. Pretrained and instruction-tuned weights are available.
Vocabulary and tool-call changes
The base card describes an expanded vocabulary. The instruction card additionally documents function calling, including tool definitions and tool-call identifiers. These capabilities belong to the Instruct variant; the base checkpoint is not automatically a tool-using assistant.Mistral 7B v0.3 official model cardMistral 7B v0.3 primary source
The matching tokenizer
The official download examples include tokenizer.model.v3. The instruction examples use that tokenizer for both chat and function calls. The base card separately identifies vocabulary expansion as its change from v0.2; function-calling behavior is documented on the Instruct card.Mistral 7B v0.3 official model cardMistral 7B v0.3 primary source
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | Mistral-7B-v0.3Mistral-7B-Instruct-v0.3 |
| Context window | 32,000 tokens The model catalog advertises 32k context for this version. This is not a measured recall limit. Mistral 7B v0.3 model and lifecycle record |
| Weight license | Apache 2.0Mistral 7B v0.3 official model card |
| Access and lifecycle | Hosted API retired on 30 March 2025. Downloadable weights remain listed.Mistral 7B v0.3 model and lifecycle recordMistral 7B v0.3 official model card |
| Checkpoint variants | Mistral-7B-v0.3 and Mistral-7B-Instruct-v0.3Mistral 7B v0.3 official model cardMistral 7B v0.3 primary source |