Sarvam AI model product
Sarvam 30B
Sarvam-30B is an open-source 30B-parameter Mixture-of-Experts reasoning model from Sarvam AI trained from scratch and optimized for Indian languages, coding, and conversational workloads.
Updated Aug 17, 2026. Default version: Sarvam-30B
LLMBoard Score33.5Sarvam-30B
Coverage80%14 benchmark families
Context window128KTokens
Official input priceN/AOfficial price unavailable
Sarvam 30B Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Sarvam 30B Benchmark Results
Benchmark scores for Sarvam-30B.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkMBPP | Score92.7% | Rank01 | Participants33 | Percentile100.0% | EvidenceC | Evaluated |
| BenchmarkBeyond AIME | Score58.3% | Rank05 | Participants5 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkHumanEval | Score92.1% | Rank08 | Participants66 | Percentile89.2% | EvidenceC | Evaluated |
| BenchmarkMATH-500 | Score97.0% | Rank09 | Participants32 | Percentile74.2% | EvidenceC | Evaluated |
| BenchmarkArena-Hard v2 | Score49.0% | Rank15 | Participants16 | Percentile6.7% | EvidenceC | Evaluated |
| BenchmarkHMMT25 | Score74.2% | Rank15 | Participants25 | Percentile41.7% | EvidenceC | Evaluated |
| BenchmarkAIME 2025 | Score96.7% | Rank18 | Participants115 | Percentile85.1% | EvidenceC | Evaluated |
| BenchmarkHMMT 2025 | Score73.3% | Rank28 | Participants33 | Percentile15.6% | EvidenceC | Evaluated |
| BenchmarkLiveCodeBench v6 | Score70.0% | Rank32 | Participants56 | Percentile43.6% | EvidenceC | Evaluated |
| BenchmarkMMLU | Score85.1% | Rank43 | Participants101 | Percentile58.0% | EvidenceC | Evaluated |
| BenchmarkBrowseComp | Score35.5% | Rank57 | Participants62 | Percentile8.2% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score80.0% | Rank59 | Participants134 | Percentile56.4% | EvidenceC | Evaluated |
| BenchmarkSWE-Bench Verified | Score34.0% | Rank105 | Participants111 | Percentile5.5% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score66.5% | Rank148 | Participants239 | Percentile38.2% | EvidenceC | Evaluated |
Sarvam 30B Arena Results
Preference and agent-evaluation results for the default version.
No Arena results
The default version does not have a matching Arena result yet.
Sarvam 30B Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.02 input, $0.10 output per 1M via FastRouter
- Tracked offerings
- 3
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderFastRouter | Provider model IDsarvam/sarvam-30b | Regionglobal | Input / 1M$0.02 | Output / 1M$0.10 | Context128K | Updated |
| ProviderNanoGPT | Provider model IDsarvam-30b | Regionglobal | Input / 1M$0.028 | Output / 1M$0.111 | Context65.5K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Sarvam 30B Runtime Performance
Provider-specific output speed and catalog latency for Sarvam-30B. Runtime does not affect the capability score.
No runtime data
Browse runtime rankingsNo provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Sarvam 30B Specifications
Technical details for the model's default version.
- Version
- Sarvam-30B
- Released
- Mar 6, 2026
- Knowledge cutoff
- Unknown
- Parameters
- 30B
- Context window
- 128K
- Max output
- 128K
- Inputs
- text
- Outputs
- text
- Open weights
- Yes
- License
- Apache 2.0
Sarvam 30B Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionSarvam-30B | Released | LLMBoard33.5 | Parameters30B | Context128K | Max output128K | Open weightsYes | LicenseApache 2.0 |
Sarvam 30B vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Sarvam 30B
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
SA
Sarvam 105B
Sarvam AI
GO
Google
DiffusionGemma 26B A4B
OP
o1 pro
OpenAI
LA
LFM2.5 2.6B
Liquid AI
MA
Mistral Small 4
Mistral AI
AC
Qwen3 VL 32B
Alibaba Cloud / Qwen Team
What is Sarvam 30B?
Key information about Sarvam 30B and its available data.
Sarvam-30B is an open-source 30B-parameter Mixture-of-Experts reasoning model from Sarvam AI trained from scratch and optimized for Indian languages, coding, and conversational workloads.
4B active parameters per token, Grouped Query Attention, and was pre-trained on 16 trillion tokens spanning code, mathematics, multilingual, and web data.
Data as of 2026-08-17.
FAQ
Common questions about Sarvam 30B.
When was Sarvam 30B released?
Sarvam 30B's default version was released on Mar 6, 2026.
How much does Sarvam 30B cost?
No official standard PAYG price is currently available for Sarvam 30B. The lowest tracked third-party offer starts at $0.02 input and $0.10 output via FastRouter.
Who created Sarvam 30B?
Sarvam 30B was created by Sarvam AI.
What is the context window for Sarvam 30B?
The default version has a 128K token context window.
Is Sarvam 30B open weight?
Yes. The default version is marked as open weight under Apache 2.0.
How many API providers offer Sarvam 30B?
3 provider offerings are linked to the default version.
What models should I compare Sarvam 30B with?
Nearby ranked alternatives include Qwen3 VL 32B, o1 pro, DiffusionGemma 26B A4B.