Sarvam M
Sarvam M is a text-only multilingual model built on Mistral Small. It supports thinking and non-thinking generation. Apache 2.0 weights remain published, but the current catalog says its API is no longer available.
Deprecated API; check access details
Use the checkpoint's thinking control
The card uses enable_thinking in the chat template and supplies different temperature guidance for the two modes. Native and romanized Indian-language input are both part of its design.Sarvam M official weights
Practical use
Practical guidance. This release can support reproducible comparisons and local adaptation. Preserve the prompt template with the model. The old API example in a model card should not override a newer service-removal notice.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Deprecated API; check access details |
| Access | Downloadable |
| Model identifiers | sarvam-m |
| Model class | 24B multilingual modelSarvam model API catalog |
| Ancestry | Post-trained Mistral Small derivativeSarvam M official weights |
Related models
- Sarvam 105B
Sarvam AI · Sarvam
A 105B mixture-of-experts reasoning model for Indian-language text, code and tool calls, with 10.3B active parameters.
Launch date not confirmedAPI · Downloadable - Sarvam 30B
Sarvam AI · Sarvam
A 30B mixture-of-experts language model with downloadable Apache 2.0 weights.
Launch date not confirmedDeprecated API; check access detailsDownloadable