Mistral 7B v0.2
Mistral 7B v0.2 changes the attention configuration and extends the advertised context from 8K to 32K. Its instruction model was introduced with the first La Plateforme API.
Three documented changes
The instruction card lists 32K context, a RoPE theta of one million and removal of sliding-window attention. It keeps the earlier INST-style conversation format. These changes concern the model and its positional encoding, not merely a different downloadable file format.Mistral 7B v0.2 official model card
What the publication date covers
The December 2023 La Plateforme launch introduced Instruct v0.2. The model catalog also links base weights. This date establishes the instruction release, not the first publication of that base archive.Mistral 7B v0.2 release announcementMistral 7B v0.2 model and lifecycle record
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | Mistral-7B-Instruct-v0.2Mistral-7B-v0.2 |
| Context window | 32,000 tokens The model catalog advertises 32k context for this version. This is not a measured recall limit. Mistral 7B v0.2 model and lifecycle record |
| Weight license | Apache 2.0Mistral 7B v0.2 official model card |
| Access and lifecycle | Hosted API retired on 30 March 2025. Downloadable weights remain listed.Mistral 7B v0.2 model and lifecycle recordMistral 7B v0.2 official model card |
| Checkpoint | Mistral-7B-Instruct-v0.2; the model catalog also links the base archiveMistral 7B v0.2 model and lifecycle record |