Sarvam 105B

high

Creator

Sarvam

Released

2026-03-06

Intelligence

11.9

Artificial Analysis Index

Coding

Artificial Analysis Index

In $/1M

$0.04

input tokens

Out $/1M

$0.17

output tokens

Blended $/1M

$0.07

3:1 blended

Speed

0

tokens / sec

Profile

License, openness, and modality — the governance layer.

License

Apache 2.0

commercial OK

Weights

Open

downloadable

Modalities

Parameters

105B

total

Sarvam-105B is Sarvam AI's flagship open-source Mixture-of-Experts reasoning model built for complex reasoning, coding, and agentic workflows. It uses 128 sparse experts with Multi-head Latent Attention for efficient long-context inference and was pre-trained on 12 trillion tokens spanning code, mathematics, multilingual, and web data.

Capability profile

Category strength across 20 domains, via LLM Stats.

legal
90finance
90language
90healthcare
90math
80biology
80physics
80chemistry
80structured output
80instruction following
80
general
70writing
70reasoning
70creativity
70code
50agents
50search
50frontend development
50vision
10
Via LLM Stats

Benchmark breakdown

Independent evaluation scores, normalized to 0–100.

GPQA Diamond
74
Humanity’s Last Exam
10
MMLU-Pro
SciCode
26
LiveCodeBench
MATH-500
AIME 2025
τ²-Bench (agentic)
47
Terminal-Bench Hard
2
IFBench
34
Via Artificial Analysis
© 2026 NYSGPT2525 LLC