Gemma 3 12B vs Mistral Small 4
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Images and charts · Long documents · Following instructions
Mistral Small 4 is the stronger all-rounder.Gemma 3 12B is cheaper.
Scores updated · 21 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.
- CodingWriting and fixing software03Mistral Small 43 of 3 tests
- ReasoningHard problems that need careful thinking02Mistral Small 42 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up02Mistral Small 42 of 2 tests
- AgentsCarrying out multi-step tasks on its own01Mistral Small 41 of 3 tests · 2 ties
- Images and chartsUnderstanding pictures, charts and video01Mistral Small 41 of 1 test
- Long documentsFinding answers in very long texts01Mistral Small 41 of 1 test
- Following instructionsDoing exactly what it is asked01Mistral Small 41 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.
Gemma 3 12B costs 73% 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 Mistral Small 4 pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+42points ahead
- Reasons across sets of long documentsAA-LCR+41.4points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+30.9points ahead
Where Gemma 3 12B pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 21 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingMistral Small 4
- SciCodeMistral Small 4 by 22.416.438.8+22.4
- Terminal-Bench 2.1Mistral Small 4 by 21021+21
- Terminal-Bench HardMistral Small 4 by 16.60.817.4+16.6
AgentsMistral Small 4
- τ-Bench V3 · BankingMistral Small 4 by 4.10.84.9+4.1
- GDPValtie00tie
- Terminal-Bench 4.0tie00tie
ReasoningMistral Small 4
- GPQA DiamondMistral Small 4 by 4234.976.9+42
- Humanity's Last ExamMistral Small 4 by 5.74.29.9+5.7
- CritPttie00.3tie
FactsMistral Small 4
- AA-Omniscience · Non-hallucinationMistral Small 4 by 30.92.633.5+30.9
- AA-Omniscience · AccuracyMistral Small 4 by 11.410.321.7+11.4
Images and chartsMistral Small 4
- MMMU-ProMistral Small 4 by 19.337.556.8+19.3
Following instructionsMistral Small 4
- IFBenchMistral Small 4 by 11.536.748.2+11.5
Other results7 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.
- AA-OmniscienceMistral Small 4 by 46.7-77.1-30.4+46.7
- LiveCodeBenchMistral Small 4 by 3924.663.6+39
- τ²-Bench Telecom (AA run)Mistral Small 4 by 30.410.841.2+30.4
- Artificial Analysis Coding IndexMistral Small 4 by 20.85.826.6+20.8
- MMLU-ProMistral Small 4 by 17.460.678+17.4
- Arena HardMistral Small 4 by 14.743.658.3+14.7
- AA IntelligenceMistral Small 4 by 7.53.811.3+7.5
Questions people ask
Which is better, Gemma 3 12B or Mistral Small 4?
Mistral Small 4 wins all seven areas where both have results: coding, agents, reasoning, facts, images and charts, long documents and following instructions. Gemma 3 12B wins none, but costs 73% less.
Which is better for coding?
Mistral Small 4. It wins 3 of the 3 coding tests both models report; Gemma 3 12B wins none.
Which is cheaper?
Gemma 3 12B costs $0.05 per million input tokens and $0.15 per million output tokens; Mistral Small 4 costs $0.15 and $0.60. That makes Gemma 3 12B about 73% cheaper for the same work.
How do you compare the two?
We use the 21 benchmark tests both models have published scores on. The verdict counts the 14 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 7 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.