Compare / Nanbeige4.3 33B vs Sarvam 30B

Nanbeige4.3 33B vs Sarvam 30B

Updated November 2026

Verdict

Nanbeige4.3 33B scores higher on the Small LM Index: 77.8 vs 61.6. On the 2 benchmarks both labs report, Nanbeige4.3 33B wins 2 and Sarvam 30B wins 0. Sarvam 30B is smaller: it needs about 20.8 GB of memory at 4-bit, against 23 GB for Nanbeige4.3 33B. Neither model has a hosted API on OpenRouter, so you would run both yourself.

Pick Nanbeige4.3 33B if…

you want the strongest model overall
you need math and reasoning
you work with long documents
you want the more widely used model (more quantized builds and community support)

Pick Sarvam 30B if…

you have limited memory (about 20.8 GB at 4-bit)

Specifications

	Nanbeige4.3 33B	Sarvam 30B
Small LM Index	77.8 (likely 77–80)	61.6 (likely 58–65)
Lab	Nanbeige (Kanzhun) · China	Sarvam AI · Rest of Asia
Released	2026-10-24	2026-03-03
Parameters	33.1B (dense)	32.2B (2.4B active, MoE)
Memory at 4-bit	~23 GB	~20.8 GB
Context window	524K tokens	131K tokens
Input	text	text
Reasoning	Always thinks	Always thinks
License	Apache-2.0	Apache-2.0
API price / 1M tokens	Self-host only	Self-host only
Hugging Face downloads (30 days)	17.8K	5K

Benchmarks

Higher is better. Bold marks the winner where both labs report a score. ≈ marks the index model's estimate for a score that wasn't reported, ✓ an independent measurement and * a score measured by another lab.

Benchmark	Nanbeige4.3 33B	Sarvam 30B
MMLU-Pro Knowledge	≈ 86	80.0
GPQA Diamond Knowledge	89.5	66.5
AIME Math	95.2	88.3
Terminal-Bench Agentic	80.9	—
τ²-Bench Agentic	—	45.7
MMLU Knowledge	—	85.1
MATH-500 Math	—	97.0
HMMT Math	—	73.3
AA-LCR Long context	84.0	—

Scores are mostly reported by each lab with its own settings, so small gaps are noise. See the methodology. Open in the interactive comparison to add more models.

Questions

Which is better, Nanbeige4.3 33B or Sarvam 30B?

Nanbeige4.3 33B scores higher on the Small LM Index: 77.8 vs 61.6. On the 2 benchmarks both labs report, Nanbeige4.3 33B wins 2 and Sarvam 30B wins 0.

Which is cheaper to run?

Neither model has a hosted API on OpenRouter, so you would run both yourself. Sarvam 30B is smaller: it needs about 20.8 GB of memory at 4-bit, against 23 GB for Nanbeige4.3 33B.

Which has the longer context window?

Nanbeige4.3 33B has the longer context window: 524K tokens against 131K.

Can I use Nanbeige4.3 33B and Sarvam 30B commercially?

Both are released under the Apache-2.0 license. Apache 2.0 and MIT allow commercial use; other licenses can add conditions, so read the license before shipping a product.