Sarvam-30B: Benchmarks, Pricing & Context Window
Sarvam-30B is a language model from Sarvam AI, released in March 2026.
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
Sarvam-30B benchmarks
Capability tiers
Standing within each category, adjusted for leaderboard depth.
Real tasks performance
High-confidence performance for Sarvam-30B across real-world prompt categories. Only 95% intervals at most 4 points wide are shown.
Performance by conversation depth
How Sarvam-30B holds up as conversations get longer.
Quality Tracker
Sarvam-30B Performance Across Datasets
Scores sourced from the model's scorecard, paper, or official blog posts
llm-stats.com - Wed Sep 16 2026Sarvam-30B model size
Sarvam-30B has 30 billion parameters and was trained on 16 trillion tokens. See how it compares to other models in the same parameter range.
Try now
huggleMake it with
Sarvam-30B.
Sarvam-30B latency
Sarvam-30B time to first token, sustained output throughput, and failed-request rate from live model usage over the trailing 7 days.
Sarvam-30B examples
Recent arena outputs from Sarvam-30B, picked from the highest-ranked matchups.
Sarvam-30B license
Sarvam-30B is released under the Apache 2.0 license, which permits commercial use, has 30.0B parameters.
- License
- Apache 2.0
- Commercial use allowed
- Parameters
- 30.0B
Apache License 2.0 - allows commercial use
Sarvam-30B resources
Official sources for Sarvam-30B: provider documentation, official playground, official launch post, model weights.
Sarvam-30B vs other models
The most-compared alternatives to Sarvam-30B are GPT-5.1 Codex High, MiMo-V2.5-Pro, Sarvam-105B. Open any pair side-by-side for benchmarks, pricing, context, and latency.
Models like Sarvam-30B
Models ranked just above and below Sarvam-30B by LLM Stats score.
FAQ
Common questions about Sarvam-30B.