Sarvam 105B vs Sarvam 30B
Wins 5 of 6 areas
Coding · Agents · Reasoning · Facts · Following instructions
Wins 0 of 6 areas
—
Sarvam 105B is the stronger all-rounder.Sarvam 30B is cheaper.
Scores updated · 24 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software30Sarvam 105B3 of 4 tests · 1 tie
- ReasoningHard problems that need careful thinking20Sarvam 105B2 of 3 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own20Sarvam 105B2 of 2 tests
- FactsGetting facts right instead of making them up20Sarvam 105B2 of 2 tests
- Following instructionsDoing exactly what it is asked10Sarvam 105B1 of 1 test
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
Long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
Sarvam 30B costs 36% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Sarvam 105B pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+14points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+11points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+10.5points ahead
Where Sarvam 30B pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 24 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingSarvam 105B
- SWE-bench VerifiedSarvam 105B by 114534+11
- SciCodeSarvam 105B by 7.226.419.2+7.2
- LiveCodeBench v6Sarvam 105B by 1.771.770+1.7
- Terminal-Bench Hardtie1.52.3tie
AgentsSarvam 105B
- BrowseCompSarvam 105B by 1449.535.5+14
- GDPValSarvam 105B by 11.911.90+11.9
ReasoningSarvam 105B
- GPQA DiamondSarvam 105B by 10.573.863.3+10.5
- Humanity's Last ExamSarvam 105B by 3.5117.5+3.5
- CritPttie00.3tie
FactsSarvam 105B
- AA-Omniscience · AccuracySarvam 105B by 517.612.6+5
- AA-Omniscience · Non-hallucinationSarvam 105B by 2.96.63.7+2.9
Following instructionsSarvam 105B
- IFBenchSarvam 105B by 7.934.426.5+7.9
Other results11 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- Arena-Hard-V2Sarvam 105B by 227149+22
- AA Agentic IndexSarvam 105B by 13.224.711.5+13.2
- τ²-Bench Telecom (AA run)Sarvam 105B by 12.346.834.5+12.3
- AA-OmniscienceSarvam 105B by 12.1-59.4-71.5+12.1
- HMMT 2025Sarvam 105B by 11.685.874.2+11.6
- BeyondAIMESarvam 105B by 10.869.158.3+10.8
- AA IntelligenceSarvam 105B by 2.28.86.6+2.2
- Artificial Analysis Coding IndexSarvam 105B by 1.99.87.9+1.9
- MMLU-ProSarvam 105B by 1.781.780+1.7
- MATH-500 (EM)Sarvam 105B by 1.698.697+1.6
- AIME 2025tie96.796.7tie
Questions people ask
Which is better, Sarvam 105B or Sarvam 30B?
Sarvam 105B wins five of the six areas where both have results: coding, agents, reasoning, facts and following instructions. Sarvam 30B wins none, but costs 36% less. They are level on long documents.
Which is better for coding?
Sarvam 105B. It wins 3 of the 4 coding tests both models report; Sarvam 30B wins none, and 1 is a tie.
Which is cheaper?
Sarvam 105B costs $0.04 per million input tokens and $0.17 per million output tokens; Sarvam 30B costs $0.03 and $0.11. That makes Sarvam 30B about 36% cheaper for the same work.
How do you compare the two?
We use the 24 benchmark tests both models have published scores on. The verdict counts the 13 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 11 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.