Claude Opus 4.5 vs Gemma 4 31B
Wins 6 of 8 areas
Coding · Agents · Reasoning · Facts · Math · Long documents
Wins 1 of 8 areas
Following instructions
Claude Opus 4.5 is the stronger all-rounder.Gemma 4 31B is cheaper and better at following instructions.
Scores updated · 51 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 software61Claude Opus 4.56 of 7 tests
- MathCompetition and research-level math30Claude Opus 4.53 of 3 tests
- FactsGetting facts right instead of making them up31Claude Opus 4.53 of 4 tests
- ReasoningHard problems that need careful thinking20Claude Opus 4.52 of 3 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own20Claude Opus 4.52 of 2 tests
- Long documentsFinding answers in very long texts10Claude Opus 4.51 of 1 test
- Following instructionsDoing exactly what it is asked01Gemma 4 31B1 of 1 test
- Images and chartsUnderstanding pictures, charts and video00Even0 each · 4 ties
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 4 31B costs 98% 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 Claude Opus 4.5 pulls ahead
- Real work tasks from 44 professionsGDPVal+41.2points ahead
- Short factual questions, answered correctlySimpleQA Verified+35.3points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+28.9points ahead
Where Gemma 4 31B pulls ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+44.7points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+17.6points ahead
Every test, side by side
All 51 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.5
- SWE-bench MultilingualGemma 4 31B by 44.7751.7+44.7
- SWE-bench VerifiedClaude Opus 4.5 by 28.980.952+28.9
- SWE-bench ProClaude Opus 4.5 by 16.35235.7+16.3
- Terminal-Bench HardClaude Opus 4.5 by 10.64736.4+10.6
- LMArena · WebDevClaude Opus 4.5 by 104 rating points14691365+104 rating
- LiveCodeBench v6Claude Opus 4.5 by 4.884.880+4.8
- SciCodeClaude Opus 4.5 by 449.545.5+4
AgentsClaude Opus 4.5
ReasoningClaude Opus 4.5
- Humanity's Last ExamClaude Opus 4.5 by 6.530.123.6+6.5
- CritPtClaude Opus 4.5 by 3.24.61.4+3.2
- GPQA Diamondtie86.685.7tie
FactsClaude Opus 4.5
- SimpleQA VerifiedClaude Opus 4.5 by 35.345.710.4+35.3
- AA-Omniscience · AccuracyClaude Opus 4.5 by 26.646.620+26.6
- AA-Omniscience · Non-hallucinationClaude Opus 4.5 by 243915+24
- Vectara HHEM hallucination ratelower is betterGemma 4 31B by 3.510.97.4+3.5
Images and chartsEven
- MathVistatie80.279.3tie
- CharXiv (RQ)tie67.267.9tie
- MMMU-Protie7473.4tie
- MMMUtie80.780.4tie
MathClaude Opus 4.5
- IMOAnswerBenchClaude Opus 4.5 by 9.58474.5+9.5
- HMMT Feb. 2026Claude Opus 4.5 by 8.185.377.2+8.1
- AIME 2026Claude Opus 4.5 by 5.995.189.2+5.9
Following instructionsGemma 4 31B
- IFBenchGemma 4 31B by 17.65875.6+17.6
Other results26 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-OmniscienceClaude Opus 4.5 by 61.914-47.9+61.9
- AA Agentic IndexClaude Opus 4.5 by 52.959.66.7+52.9
- QwenWebBenchClaude Opus 4.5 by 339 rating points15361197+339 rating
- τ²-Bench Telecom (AA run)Claude Opus 4.5 by 29.689.559.9+29.6
- Claw-Eval AvgClaude Opus 4.5 by 28.176.648.5+28.1
- NL2RepoClaude Opus 4.5 by 27.743.215.5+27.7
- SkillsBench Avg5Claude Opus 4.5 by 21.745.323.6+21.7
- τ²-BenchClaude Opus 4.5 by 21.398.276.9+21.3
- SimpleVQAClaude Opus 4.5 by 16.869.752.9+16.8
- HLE (with tools)Claude Opus 4.5 by 16.743.226.5+16.7
- Terminal-Bench 2.0Claude Opus 4.5 by 16.459.342.9+16.4
- AA IntelligenceClaude Opus 4.5 by 14.429.114.7+14.4
- QwenClawBenchClaude Opus 4.5 by 10.652.341.7+10.6
- C-EvalClaude Opus 4.5 by 9.692.282.6+9.6
- MATH-VisionGemma 4 31B by 8.577.185.6+8.5
- OmniDocBench 1.5Claude Opus 4.5 by 7.687.780.1+7.6
- HMMT Nov. 2025Claude Opus 4.5 by 5.893.387.5+5.8
- SuperGPQAClaude Opus 4.5 by 4.970.665.7+4.9
- Artificial Analysis Coding IndexClaude Opus 4.5 by 4.447.843.4+4.4
- MMLU-ProClaude Opus 4.5 by 4.389.585.2+4.3
- HMMT Feb. 2025Claude Opus 4.5 by 4.292.988.7+4.2
- vectara_avg_summary_lengthClaude Opus 4.5 by 38.7 rating points114.575.8+38.7 rating
- vectara_factual_consistencyGemma 4 31B by 3.589.192.6+3.5
- MMMLUClaude Opus 4.5 by 2.490.888.4+2.4
- MMLU-ReduxClaude Opus 4.5 by 1.995.693.7+1.9
- vectara_answer_rateGemma 4 31B by 1.398.7100+1.3
Questions people ask
Which is better, Claude Opus 4.5 or Gemma 4 31B?
Claude Opus 4.5 wins six of the eight areas we test: coding, agents, reasoning, facts, math and long documents. Gemma 4 31B wins following instructions, and costs 98% less. They are level on images and charts.
Which is better for coding?
Claude Opus 4.5. It wins 6 of the 7 coding tests both models report; Gemma 4 31B wins 1.
Which is cheaper?
Claude Opus 4.5 costs $5.00 per million input tokens and $25.00 per million output tokens; Gemma 4 31B costs $0.17 and $0.40. That makes Gemma 4 31B about 98% cheaper for the same work.
How do you compare the two?
We use the 51 benchmark tests both models have published scores on. The verdict counts the 25 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 26 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.