Devstral Small 2 vs Gemma 4 31B
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Images and charts · Long documents · Following instructions
Gemma 4 31B is the stronger all-rounder.
Scores updated · 22 tests both models report · How we compare
Where each one wins
Tests won in each of the seven 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.
- ReasoningHard problems that need careful thinking03Gemma 4 31B3 of 3 tests
- AgentsCarrying out multi-step tasks on its own02Gemma 4 31B2 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up02Gemma 4 31B2 of 2 tests
- CodingWriting and fixing software23Gemma 4 31B3 of 5 tests
- Images and chartsUnderstanding pictures, charts and video01Gemma 4 31B1 of 1 test
- Long documentsFinding answers in very long texts01Gemma 4 31B1 of 1 test
- Following instructionsDoing exactly what it is asked01Gemma 4 31B1 of 1 test
Images and charts, 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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Gemma 4 31B pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+44.4points ahead
- Reasons across sets of long documentsAA-LCR+41.7points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+32.5points ahead
Where Devstral Small 2 pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+16points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+4points ahead
Every test, side by side
All 22 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemma 4 31B
- Terminal-Bench HardGemma 4 31B by 19.716.736.4+19.7
- SWE-bench VerifiedDevstral Small 2 by 166852+16
- Terminal-Bench 2.1Gemma 4 31B by 13.829.643.4+13.8
- SciCodeGemma 4 31B by 13.132.445.5+13.1
- SWE-bench MultilingualDevstral Small 2 by 455.751.7+4
AgentsGemma 4 31B
- GDPValGemma 4 31B by 4.51.66.1+4.5
- τ-Bench V3 · BankingGemma 4 31B by 4.110.714.8+4.1
- Terminal-Bench 4.0tie00tie
ReasoningGemma 4 31B
- GPQA DiamondGemma 4 31B by 32.553.285.7+32.5
- Humanity's Last ExamGemma 4 31B by 20.13.523.6+20.1
- CritPtGemma 4 31B by 1.401.4+1.4
FactsGemma 4 31B
- AA-Omniscience · AccuracyGemma 4 31B by 4.215.820+4.2
- AA-Omniscience · Non-hallucinationGemma 4 31B by 2.312.715+2.3
Following instructionsGemma 4 31B
- IFBenchGemma 4 31B by 44.431.275.6+44.4
Other results6 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.
- τ²-Bench Telecom (AA run)Gemma 4 31B by 36.523.459.9+36.5
- Terminal-Bench 2.0Gemma 4 31B by 20.422.542.9+20.4
- Artificial Analysis Coding IndexGemma 4 31B by 14.129.343.4+14.1
- AA-OmniscienceGemma 4 31B by 9.8-57.7-47.9+9.8
- AA IntelligenceGemma 4 31B by 7.27.514.7+7.2
- AA Agentic IndexGemma 4 31B by 1.94.86.7+1.9
Questions people ask
Which is better, Devstral Small 2 or Gemma 4 31B?
Gemma 4 31B wins all seven areas where both have results: coding, agents, reasoning, facts, images and charts, long documents and following instructions. Devstral Small 2 wins none.
Which is better for coding?
Gemma 4 31B. It wins 3 of the 5 coding tests both models report; Devstral Small 2 wins 2.
How do you compare the two?
We use the 22 benchmark tests both models have published scores on. The verdict counts the 16 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 6 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.