Gemma 4 12B vs Gemma 4 31B
Wins 0 of 8 areas
—
Wins 7 of 8 areas
Coding · Agents · Reasoning · Images and charts · Math · Long documents · Following instructions
Gemma 4 31B is the stronger all-rounder.Gemma 4 12B is cheaper.
Scores updated · 30 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. 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 software04Gemma 4 31B4 of 4 tests
- ReasoningHard problems that need careful thinking03Gemma 4 31B3 of 3 tests
- AgentsCarrying out multi-step tasks on its own02Gemma 4 31B2 of 2 tests
- Long documentsFinding answers in very long texts02Gemma 4 31B2 of 2 tests
- Images and chartsUnderstanding pictures, charts and video01Gemma 4 31B1 of 1 test
- MathCompetition and research-level math01Gemma 4 31B1 of 1 test
- Following instructionsDoing exactly what it is asked01Gemma 4 31B1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Images and charts, math 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 4 12B costs 29% 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 Gemma 4 31B pulls ahead
- Finds one of eight look-alike replies in a long chatMRCR v2 8 needle 128k (average)+23points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+18.2points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+16.1points ahead
Where Gemma 4 12B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+4points ahead
Every test, side by side
All 30 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 18.218.236.4+18.2
- Terminal-Bench 2.1Gemma 4 31B by 16.127.343.4+16.1
- LiveCodeBench v6Gemma 4 31B by 87280+8
- SciCodeGemma 4 31B by 7.338.245.5+7.3
AgentsGemma 4 31B
- GDPValGemma 4 31B by 6.106.1+6.1
- τ-Bench V3 · BankingGemma 4 31B by 6.18.714.8+6.1
ReasoningGemma 4 31B
- GPQA DiamondGemma 4 31B by 10.475.385.7+10.4
- Humanity's Last ExamGemma 4 31B by 7.915.723.6+7.9
- CritPtGemma 4 31B by 1.401.4+1.4
FactsEven
- AA-Omniscience · AccuracyGemma 4 31B by 4.415.620+4.4
- AA-Omniscience · Non-hallucinationGemma 4 12B by 41915+4
Long documentsGemma 4 31B
- MRCR v2 8 needle 128k (average)Gemma 4 31B by 2343.466.4+23
- AA-LCRGemma 4 31B by 663.769.7+6
Following instructionsGemma 4 31B
- IFBenchGemma 4 31B by 2.173.575.6+2.1
Other results14 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.
- Codeforces ELOGemma 4 31B by 491 rating points16592150+491 rating
- τ²-Bench Telecom (AA run)Gemma 4 31B by 23.636.359.9+23.6
- BIG-Bench Extra HardGemma 4 31B by 21.45374.4+21.4
- HLE (with tools)Gemma 4 31B by 21.35.226.5+21.3
- MedXPertQA MMGemma 4 31B by 12.648.761.3+12.6
- Artificial Analysis Coding IndexGemma 4 31B by 12.43143.4+12.4
- MMLU-ProGemma 4 31B by 877.285.2+8
- τ²-BenchGemma 4 31B by 7.96976.9+7.9
- MATH-VisionGemma 4 31B by 5.979.785.6+5.9
- MMMLUGemma 4 31B by 583.488.4+5
- AA-OmniscienceGemma 4 31B by 4.8-52.7-47.9+4.8
- AA Agentic IndexGemma 4 12B by 1.27.96.7+1.2
- AA Intelligencetie14.214.7tie
- OmniDocBench 1.5 (average edit distance, lower is better)lower is bettertie0.20.1tie
Questions people ask
Which is better, Gemma 4 12B or Gemma 4 31B?
Gemma 4 31B wins seven of the eight areas we test: coding, agents, reasoning, images and charts, math, long documents and following instructions. Gemma 4 12B wins none, but costs 29% less. They are level on facts.
Which is better for coding?
Gemma 4 31B. It wins 4 of the 4 coding tests both models report; Gemma 4 12B wins none.
Which is cheaper?
Gemma 4 12B costs $0.10 per million input tokens and $0.30 per million output tokens; Gemma 4 31B costs $0.17 and $0.40. That makes Gemma 4 12B about 29% cheaper for the same work.
How do you compare the two?
We use the 30 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 14 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.