GPT-5.4 vs Qwen3.6 Plus
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents
Wins 0 of 8 areas
—
GPT-5.4 is the stronger all-rounder.Qwen3.6 Plus is cheaper.
Scores updated · 57 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.
- MathCompetition and research-level math50GPT-5.45 of 5 tests
- CodingWriting and fixing software51GPT-5.45 of 8 tests · 2 ties
- ReasoningHard problems that need careful thinking40GPT-5.44 of 4 tests
- AgentsCarrying out multi-step tasks on its own31GPT-5.43 of 4 tests
- FactsGetting facts right instead of making them up21GPT-5.42 of 3 tests
- Images and chartsUnderstanding pictures, charts and video10GPT-5.41 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10GPT-5.41 of 1 test
- Following instructionsDoing exactly what it is asked11Even1 each
Long documents rests on a single test.
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.
Qwen3.6 Plus costs 80% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.4 pulls ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+40.3points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+24.5points ahead
- Research-level physics problemsCritPt+20.5points ahead
Where Qwen3.6 Plus pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+57.1points ahead
- Multi-step tasks using many real software toolsMCP Atlas+6.9points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+2.1points ahead
Every test, side by side
All 57 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.4
- Terminal-Bench 2.1GPT-5.4 by 16.978.361.4+16.9
- SciCodeGPT-5.4 by 15.956.640.7+15.9
- Terminal-Bench HardGPT-5.4 by 13.757.643.9+13.7
- LiveBench · Agentic CodingGPT-5.4 by 12.453.841.4+12.4
- SWE-bench MultilingualQwen3.6 Plus by 2.171.773.8+2.1
- SWE-bench ProGPT-5.4 by 1.157.756.6+1.1
- LiveBench · Codingtie77.578.2tie
- LMArena · WebDevtie14651461tie
AgentsGPT-5.4
- τ-Bench V3 · BankingGPT-5.4 by 18.839.620.8+18.8
- GDPValGPT-5.4 by 12.737.424.7+12.7
- OSWorld-VerifiedGPT-5.4 by 12.57562.5+12.5
- MCP AtlasQwen3.6 Plus by 6.967.274.1+6.9
ReasoningGPT-5.4
- CritPtGPT-5.4 by 20.523.42.9+20.5
- Humanity's Last ExamGPT-5.4 by 15.943.727.8+15.9
- LiveBench · ReasoningGPT-5.4 by 12.388.175.8+12.3
- GPQA DiamondGPT-5.4 by 3.89288.2+3.8
FactsGPT-5.4
- AA-Omniscience · Non-hallucinationQwen3.6 Plus by 57.18.365.4+57.1
- AA-Omniscience · AccuracyGPT-5.4 by 24.550.926.4+24.5
- SimpleQA VerifiedGPT-5.4 by 145.144.1+1
Images and chartsGPT-5.4
- CharXiv (RQ)GPT-5.4 by 1.382.881.5+1.3
- MMMU-Protie78.478tie
MathGPT-5.4
- FrontierMath Tiers 1-3 (v2)GPT-5.4 by 40.378.638.3+40.3
- LiveBench · MathematicsGPT-5.4 by 10.594.283.7+10.5
- HMMT Feb. 2026GPT-5.4 by 9.997.787.8+9.9
- IMOAnswerBenchGPT-5.4 by 7.691.483.8+7.6
- AIME 2026GPT-5.4 by 3.999.295.3+3.9
Following instructionsEven
- LiveBench · Instruction FollowingGPT-5.4 by 11.970.258.3+11.9
- IFBenchQwen3.6 Plus by 1.373.975.2+1.3
Other results28 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.
- frontiermath_tier_4_v1GPT-5.4 by 18.827.18.3+18.8
- Artificial Analysis Coding IndexGPT-5.4 by 16.671.154.5+16.6
- AgentWorldBench - SearchGPT-5.4 by 15.437.321.9+15.4
- AA Agentic IndexGPT-5.4 by 15.244.229+15.2
- AgentWorldBench - MCPGPT-5.4 by 14.870.155.3+14.8
- Tool-DecathlonGPT-5.4 by 14.854.639.8+14.8
- ToolathlonGPT-5.4 by 14.854.639.8+14.8
- MCPMarkGPT-5.4 by 14.362.548.2+14.3
- Terminal-Bench 2.0GPT-5.4 by 13.575.161.6+13.5
- AA IntelligenceGPT-5.4 by 123927+12
- τ²-Bench Telecom (AA run)Qwen3.6 Plus by 10.687.197.7+10.6
- LiveBenchGPT-5.4 by 9.480.370.8+9.4
- livebench_data_analysisGPT-5.4 by 9.479.369.9+9.4
- AgentWorldBench - OSGPT-5.4 by 8.368.660.3+8.3
- livebench_languageGPT-5.4 by 7.682.675+7.6
- AgentWorldBench - OverallGPT-5.4 by 7.558.350.8+7.5
- AgentWorldBench - SWEGPT-5.4 by 7.266.359.1+7.2
- AA-OmniscienceGPT-5.4 by 4.95.80.9+4.9
- MATH-VisionGPT-5.4 by 49288+4
- NL2RepoGPT-5.4 by 3.441.337.9+3.4
- AgentWorldBench - TerminalGPT-5.4 by 3.153.750.6+3.1
- AgentWorldBench - AndroidGPT-5.4 by 2.36057.6+2.3
- τ³-BenchGPT-5.4 by 2.272.970.7+2.2
- OmniDocBench 1.5Qwen3.6 Plus by 2.189.191.2+2.1
- HLE (with tools)GPT-5.4 by 1.45250.6+1.4
- HMMT Nov. 2025GPT-5.4 by 1.295.894.6+1.2
- AgentWorldBench - WebGPT-5.4 by 151.850.8+1
- MMLU-ProQwen3.6 Plus by 187.588.5+1
Questions people ask
Which is better, GPT-5.4 or Qwen3.6 Plus?
GPT-5.4 wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and long documents. Qwen3.6 Plus wins none, but costs 80% less. They are level on following instructions.
Which is better for coding?
GPT-5.4. It wins 5 of the 8 coding tests both models report; Qwen3.6 Plus wins 1, and 2 are ties.
Which is cheaper?
GPT-5.4 costs $2.50 per million input tokens and $15.00 per million output tokens; Qwen3.6 Plus costs $0.50 and $3.00. That makes Qwen3.6 Plus about 80% cheaper for the same work.
How do you compare the two?
We use the 57 benchmark tests both models have published scores on. The verdict counts the 29 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 28 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.