Sarvam Models by Sarvam.ai
Last refreshed 2026-11-09. Next refresh: weekly.
Details
Capabilities
About
Sarvam is Sarvam AI’s Indian-language model line. Sarvam-M is a 24B multilingual hybrid-reasoning text model post-trained from Mistral Small for Indic languages plus math and programming, with both “think” and “non-think” modes, support for Indic scripts and romanized forms, and character training toward Indian cultural context. Sarvam presents it as part of building a sovereign AI ecosystem in India with regular model drops and technical write-ups.
Decision facts
- Best fit
- classificationreasoningJSON / Tool use
- Capability starting point
- Sarvam Prism with 128K context and structured outputs
- Lowest tracked input
- Not tracked
- Closest related family
- 360Zhinao
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
Use when the workload needs classification, 128K context, and structured outputs.
Use when the workload needs reasoning, 65.5K context, and JSON / Tool use.
Use when the workload needs 128K context and 24B parameters.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Sarvam Prism | Use when the workload needs classification, 128K context, and structured outputs. | 2026-11 | classification128K contextstructured outputs | Current |
| Sarvam 30B | Use when the workload needs reasoning, 65.5K context, and JSON / Tool use. | 2026-03 | reasoning65.5K contextJSON / Tool use | Current |
| Sarvam-M Multilingual Hybrid | Use when the workload needs 128K context and 24B parameters. | 2025-06 | 128K context24B parameters | Current |
Release Timeline
3 release groupsSpecifications(3 models)
| Model | Released | Context | Parameters | JSON / Tool use | Structured Outputs |
|---|---|---|---|---|---|
| Sarvam Prism | 2026-11 | 128K | 105B (10B active) | No | Yes |
| Sarvam 30B | 2026-03 | 65.5K | 30B (2.4B active) | Yes | No |
| Sarvam-M Multilingual Hybrid | 2025-06 | 128K | 24B | No | No |
Available From(2 providers)
Popular comparisons in this family
- Llama 2 7B vs Sarvam-M Multilingual Hybrid25
- Claude Instant vs Sarvam 30B23
- Mistral Small 3 vs Sarvam-M Multilingual Hybrid20
- Hunyuan Hy3 Preview vs Sarvam-M Multilingual Hybrid16
- Claude Instant vs Sarvam-M Multilingual Hybrid13
- Mistral Large 3 675B Instruct vs Sarvam-M Multilingual Hybrid11
- Mistral 7B v0.1 vs Sarvam-M Multilingual Hybrid10
- GPT-4 Turbo vs Sarvam 30B10
- Together AI - Llama 3 8B Lite vs Sarvam-M Multilingual Hybrid10
- Llama 3.1 Swallow 8B Instruct vs Sarvam-M Multilingual Hybrid9