Gemma 4 31B vs Qwen3.6 Plus
Wins 0 of 8 areas
—
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents
Qwen3.6 Plus is the stronger all-rounder.Gemma 4 31B is cheaper.
Scores updated · 52 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 software17Qwen3.6 Plus7 of 8 tests
- AgentsCarrying out multi-step tasks on its own03Qwen3.6 Plus3 of 3 tests
- ReasoningHard problems that need careful thinking03Qwen3.6 Plus3 of 3 tests
- FactsGetting facts right instead of making them up03Qwen3.6 Plus3 of 3 tests
- Images and chartsUnderstanding pictures, charts and video03Qwen3.6 Plus3 of 3 tests
- MathCompetition and research-level math03Qwen3.6 Plus3 of 3 tests
- Long documentsFinding answers in very long texts01Qwen3.6 Plus1 of 1 test
- Following instructionsDoing exactly what it is asked00Even0 each · 1 tie
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 84% 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 Qwen3.6 Plus pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+50.4points ahead
- Short factual questions, answered correctlySimpleQA Verified+33.7points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+26.8points ahead
Where Gemma 4 31B pulls ahead
- Code for real scientific research problemsSciCode+4.8points ahead
Every test, side by side
All 52 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.6 Plus
- SWE-bench VerifiedQwen3.6 Plus by 26.85278.8+26.8
- SWE-bench MultilingualQwen3.6 Plus by 22.151.773.8+22.1
- SWE-bench ProQwen3.6 Plus by 20.935.756.6+20.9
- Terminal-Bench 2.1Qwen3.6 Plus by 1843.461.4+18
- LMArena · WebDevQwen3.6 Plus by 96 rating points13651461+96 rating
- Terminal-Bench HardQwen3.6 Plus by 7.536.443.9+7.5
- LiveCodeBench v6Qwen3.6 Plus by 7.18087.1+7.1
- SciCodeGemma 4 31B by 4.845.540.7+4.8
AgentsQwen3.6 Plus
- GDPValQwen3.6 Plus by 18.66.124.7+18.6
- MCP AtlasQwen3.6 Plus by 16.957.274.1+16.9
- τ-Bench V3 · BankingQwen3.6 Plus by 614.820.8+6
ReasoningQwen3.6 Plus
- Humanity's Last ExamQwen3.6 Plus by 4.223.627.8+4.2
- GPQA DiamondQwen3.6 Plus by 2.585.788.2+2.5
- CritPtQwen3.6 Plus by 1.51.42.9+1.5
FactsQwen3.6 Plus
- AA-Omniscience · Non-hallucinationQwen3.6 Plus by 50.41565.4+50.4
- SimpleQA VerifiedQwen3.6 Plus by 33.710.444.1+33.7
- AA-Omniscience · AccuracyQwen3.6 Plus by 6.42026.4+6.4
Images and chartsQwen3.6 Plus
- CharXiv (RQ)Qwen3.6 Plus by 13.667.981.5+13.6
- MMMUQwen3.6 Plus by 5.680.486+5.6
- MMMU-ProQwen3.6 Plus by 4.673.478+4.6
MathQwen3.6 Plus
- HMMT Feb. 2026Qwen3.6 Plus by 10.677.287.8+10.6
- IMOAnswerBenchQwen3.6 Plus by 9.374.583.8+9.3
- AIME 2026Qwen3.6 Plus by 6.189.295.3+6.1
Other results27 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-OmniscienceQwen3.6 Plus by 48.8-47.90.9+48.8
- WideSearchQwen3.6 Plus by 39.135.274.3+39.1
- τ²-Bench Telecom (AA run)Qwen3.6 Plus by 37.859.997.7+37.8
- Claw Eval (pass@3)Qwen3.6 Plus by 33.72558.7+33.7
- MCPMarkQwen3.6 Plus by 30.118.148.2+30.1
- HLE (with tools)Qwen3.6 Plus by 24.126.550.6+24.1
- NL2RepoQwen3.6 Plus by 22.415.537.9+22.4
- AA Agentic IndexQwen3.6 Plus by 22.36.729+22.3
- Terminal-Bench 2.0Qwen3.6 Plus by 18.742.961.6+18.7
- Tool-DecathlonQwen3.6 Plus by 18.621.239.8+18.6
- DeepPlanningQwen3.6 Plus by 17.52441.5+17.5
- SimpleVQAQwen3.6 Plus by 14.452.967.3+14.4
- RealWorldQAQwen3.6 Plus by 13.172.385.4+13.1
- AA IntelligenceQwen3.6 Plus by 12.314.727+12.3
- Artificial Analysis Coding IndexQwen3.6 Plus by 11.143.454.5+11.1
- OmniDocBench 1.5Qwen3.6 Plus by 11.180.191.2+11.1
- C-EvalQwen3.6 Plus by 10.782.693.3+10.7
- CC-OCRQwen3.6 Plus by 7.775.783.4+7.7
- HMMT Nov. 2025Qwen3.6 Plus by 7.187.594.6+7.1
- SuperGPQAQwen3.6 Plus by 5.965.771.6+5.9
- AI2DQwen3.6 Plus by 5.48994.4+5.4
- MMLU-ProQwen3.6 Plus by 3.385.288.5+3.3
- τ³-BenchQwen3.6 Plus by 3.267.570.7+3.2
- MATH-VisionQwen3.6 Plus by 2.485.688+2.4
- VITA-BenchQwen3.6 Plus by 1.34344.3+1.3
- MMMLUQwen3.6 Plus by 1.188.489.5+1.1
- MMLU-Reduxtie93.794.5tie
Questions people ask
Which is better, Gemma 4 31B or Qwen3.6 Plus?
Qwen3.6 Plus wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and long documents. Gemma 4 31B wins none, but costs 84% less. They are level on following instructions.
Which is better for coding?
Qwen3.6 Plus. It wins 7 of the 8 coding tests both models report; Gemma 4 31B wins 1.
Which is cheaper?
Gemma 4 31B costs $0.17 per million input tokens and $0.40 per million output tokens; Qwen3.6 Plus costs $0.50 and $3.00. That makes Gemma 4 31B about 84% cheaper for the same work.
How do you compare the two?
We use the 52 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 27 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.