GPT-5.4 vs Qwen3.7 Max
Wins 4 of 7 areas
Coding · Reasoning · Math · Long documents
Wins 2 of 7 areas
Facts · Following instructions
GPT-5.4 wins more areas, narrowly.Qwen3.7 Max is cheaper and better at following instructions.
Scores updated · 43 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. 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 6 tests · 1 tie
- ReasoningHard problems that need careful thinking30GPT-5.43 of 4 tests · 1 tie
- CodingWriting and fixing software53GPT-5.45 of 8 tests
- Long documentsFinding answers in very long texts10GPT-5.41 of 1 test
- Following instructionsDoing exactly what it is asked02Qwen3.7 Max2 of 2 tests
- FactsGetting facts right instead of making them up12Qwen3.7 Max2 of 3 tests
- AgentsCarrying out multi-step tasks on its own22Even2 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.
Qwen3.7 Max costs 43% 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
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+27.8points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+19.8points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+14.8points ahead
Where Qwen3.7 Max pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+66.1points ahead
- Short factual questions, answered correctlySimpleQA Verified+10.7points ahead
- Multi-step tasks using many real software toolsMCP Atlas+9.2points ahead
Every test, side by side
All 43 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.4
- LiveBench · Agentic CodingGPT-5.4 by 10.253.843.6+10.2
- SciCodeGPT-5.4 by 7.156.649.5+7.1
- Terminal-Bench HardGPT-5.4 by 6.857.650.8+6.8
- SWE-bench MultilingualQwen3.7 Max by 6.671.778.3+6.6
- LMArena · WebDevQwen3.7 Max by 50 rating points14651515+50 rating
- Terminal-Bench 2.1GPT-5.4 by 3.878.374.5+3.8
- LiveBench · CodingGPT-5.4 by 3.377.574.2+3.3
- SWE-bench ProQwen3.7 Max by 2.957.760.6+2.9
AgentsEven
- τ-Bench V3 · BankingGPT-5.4 by 27.839.611.8+27.8
- MCP AtlasQwen3.7 Max by 9.267.276.4+9.2
- AA IT-Bench SREQwen3.7 Max by 834.542.5+8
- GDPValGPT-5.4 by 5.837.431.6+5.8
ReasoningGPT-5.4
- CritPtGPT-5.4 by 1023.413.4+10
- LiveBench · ReasoningGPT-5.4 by 4.888.183.3+4.8
- Humanity's Last ExamGPT-5.4 by 3.243.740.5+3.2
- GPQA Diamondtie9292.3tie
FactsQwen3.7 Max
- AA-Omniscience · Non-hallucinationQwen3.7 Max by 66.18.374.4+66.1
- AA-Omniscience · AccuracyGPT-5.4 by 19.850.931.1+19.8
- SimpleQA VerifiedQwen3.7 Max by 10.745.155.8+10.7
MathGPT-5.4
- FrontierMath Tier 4GPT-5.4 by 14.84934.1+14.8
- FrontierMath Tiers 1-3 (v2)GPT-5.4 by 1478.664.6+14
- LiveBench · MathematicsGPT-5.4 by 8.994.285.3+8.9
- AIME 2026GPT-5.4 by 2.299.297+2.2
- IMOAnswerBenchGPT-5.4 by 1.491.490+1.4
- HMMT Feb. 2026tie97.797.1tie
Following instructionsQwen3.7 Max
- IFBenchQwen3.7 Max by 6.673.980.5+6.6
- LiveBench · Instruction FollowingQwen3.7 Max by 3.870.274+3.8
Other results15 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.
- DeepSWE 1.1GPT-5.4 by 30.45221.6+30.4
- AA Agentic IndexGPT-5.4 by 20.344.223.9+20.3
- AA IntelligenceGPT-5.4 by 9.53929.5+9.5
- AA-OmniscienceQwen3.7 Max by 7.75.813.5+7.7
- τ²-Bench Telecom (AA run)Qwen3.7 Max by 7.687.194.7+7.6
- livebench_data_analysisGPT-5.4 by 7.579.371.8+7.5
- LiveBenchGPT-5.4 by 680.374.3+6
- NL2RepoQwen3.7 Max by 5.941.347.2+5.9
- Terminal-Bench 2.0GPT-5.4 by 5.475.169.7+5.4
- Artificial Analysis Coding IndexGPT-5.4 by 5.171.166+5.1
- livebench_languageGPT-5.4 by 2.982.679.7+2.9
- MMLU-ProQwen3.7 Max by 2.187.589.6+2.1
- MCPMarkGPT-5.4 by 1.762.560.8+1.7
- HLE (with tools)Qwen3.7 Max by 1.452.153.5+1.4
- HMMT Nov. 2025tie95.895tie
Questions people ask
Which is better, GPT-5.4 or Qwen3.7 Max?
GPT-5.4 wins four of the seven areas where both have results: coding, reasoning, math and long documents. Qwen3.7 Max wins facts and following instructions, and costs 43% less. They are level on agents.
Which is better for coding?
GPT-5.4. It wins 5 of the 8 coding tests both models report; Qwen3.7 Max wins 3.
Which is cheaper?
GPT-5.4 costs $2.50 per million input tokens and $15.00 per million output tokens; Qwen3.7 Max costs $2.50 and $7.50. That makes Qwen3.7 Max about 43% cheaper for the same work.
How do you compare the two?
We use the 43 benchmark tests both models have published scores on. The verdict counts the 28 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 15 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.