Gemma 4 31B vs Gemma 4 E2B
Wins 7 of 8 areas
Coding · Agents · Reasoning · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
Gemma 4 31B is the stronger all-rounder.
Scores updated · 34 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 software30Gemma 4 31B3 of 3 tests
- ReasoningHard problems that need careful thinking30Gemma 4 31B3 of 3 tests
- AgentsCarrying out multi-step tasks on its own20Gemma 4 31B2 of 2 tests
- Images and chartsUnderstanding pictures, charts and video20Gemma 4 31B2 of 2 tests
- Long documentsFinding answers in very long texts20Gemma 4 31B2 of 2 tests
- MathCompetition and research-level math10Gemma 4 31B1 of 1 test
- Following instructionsDoing exactly what it is asked10Gemma 4 31B1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
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.
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
- Reasons across sets of long documentsAA-LCR+53.4points ahead
- US invitational high-school math exam problemsAIME 2026+51.7points ahead
- Finds one of eight look-alike replies in a long chatMRCR v2 8 needle 128k (average)+47.3points ahead
Where Gemma 4 E2B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+52.6points ahead
Every test, side by side
All 34 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemma 4 31B
- LiveCodeBench v6Gemma 4 31B by 368044+36
- Terminal-Bench HardGemma 4 31B by 33.436.43+33.4
- SciCodeGemma 4 31B by 24.645.520.9+24.6
AgentsGemma 4 31B
- τ-Bench V3 · BankingGemma 4 31B by 10.314.84.5+10.3
- GDPValGemma 4 31B by 6.16.10+6.1
ReasoningGemma 4 31B
- GPQA DiamondGemma 4 31B by 42.485.743.3+42.4
- Humanity's Last ExamGemma 4 31B by 18.823.64.8+18.8
- CritPtGemma 4 31B by 1.41.40+1.4
FactsEven
- AA-Omniscience · Non-hallucinationGemma 4 E2B by 52.61567.6+52.6
- AA-Omniscience · AccuracyGemma 4 31B by 13.4206.6+13.4
Images and chartsGemma 4 31B
Long documentsGemma 4 31B
- AA-LCRGemma 4 31B by 53.469.716.3+53.4
- MRCR v2 8 needle 128k (average)Gemma 4 31B by 47.366.419.1+47.3
Following instructionsGemma 4 31B
- IFBenchGemma 4 31B by 37.675.638+37.6
Other results18 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 1517 rating points2150633+1517 rating
- t2-benchGemma 4 31B by 5786.429.4+57
- BIG-Bench Extra HardGemma 4 31B by 52.574.421.9+52.5
- τ²-BenchGemma 4 31B by 52.476.924.5+52.4
- τ²-Bench Telecom (AA run)Gemma 4 31B by 39.159.920.8+39.1
- MedXPertQA MMGemma 4 31B by 37.861.323.5+37.8
- Artificial Analysis Coding IndexGemma 4 31B by 36.243.47.2+36.2
- MATH-VisionGemma 4 31B by 33.285.652.4+33.2
- SimpleVQAGemma 4 31B by 25.652.927.3+25.6
- MMLU-ProGemma 4 31B by 25.285.260+25.2
- AA-OmniscienceGemma 4 E2B by 24.3-47.9-23.6+24.3
- MMMLUGemma 4 31B by 2188.467.4+21
- RealWorldQAGemma 4 31B by 12.372.360+12.3
- HallusionBenchGemma 4 31B by 7.667.459.8+7.6
- AA IntelligenceGemma 4 31B by 6.914.77.8+6.9
- AA Agentic IndexGemma 4 31B by 5.26.71.5+5.2
- Claw-Eval AvgGemma 4 E2B by 4.648.553.1+4.6
- OmniDocBench 1.5 (average edit distance, lower is better)lower is bettertie0.10.3tie
Questions people ask
Which is better, Gemma 4 31B or Gemma 4 E2B?
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 E2B wins none. They are level on facts.
Which is better for coding?
Gemma 4 31B. It wins 3 of the 3 coding tests both models report; Gemma 4 E2B wins none.
How do you compare the two?
We use the 34 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 18 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.