Sarvam 30B (High)
SarvamOpen WeightApache 2.0 · Commercial OK
Description
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. It uses 128 sparse experts with 2.4B active parameters per token, Grouped Query Attention, and was pre-trained on 16 trillion tokens spanning code, mathematics, multilingual, and web data.
Release Date
2026-03-06
Parameters
30.0B
Context Length
66K
Modalities
text
Capability Radar
70
general
60
coding
80
reasoning
60
scienceest.
40
agents
0
multimodal
Science is estimated from LLM Stats science scores or reasoning when no dedicated science benchmarks are available.
Rankings
No ranking data available
Benchmark Scores (LLM Stats)
(LLM Stats (zeroeval))General
MMLU
85.1%SR
Arena-Hard v2
49.0%SR
Language
MMLU-Pro
80.0%SR
Math
MATH-500
97.0%SR
AIME 2025
96.7%SR
HMMT25
74.2%SR
HMMT 2025
73.3%SR
Beyond AIME
58.3%SR
Reasoning
MBPP
0.93 / 100SR
HumanEvalOpenAI (2021)
92.1%SR
LiveCodeBench v6
70.0%SR
GPQANYU + Cohere + Anthropic (2023)
66.5%SR
BrowseCompOpenAI (2025)
35.5%SR
SWE-Bench Verified
34.0%SR
AA Evaluation Indices
(Artificial Analysis)No AA evaluation data available
LLM Stats Category Scores
(LLM Stats (zeroeval))Language80
Legal80
Math80
Finance80
Healthcare80
Physics70
Reasoning70
General70
Biology70
Chemistry70
Code60
Creativity50
Writing50
Search40
Agents40
Frontend Development30
Pricing
No pricing data available
Speed
No speed data available
Provider Price Ranking
No provider data available