GPT-5.4 vs Qwen3.8 Max Preview
Wins 3 of 8 areas
Reasoning · Math · Long documents
Wins 5 of 8 areas
Coding · Agents · Facts · Images and charts · Following instructions
Qwen3.8 Max Preview is the stronger all-rounder.GPT-5.4 is better at math.
Scores updated · 37 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.
- AgentsCarrying out multi-step tasks on its own05Qwen3.8 Max Preview5 of 5 tests
- CodingWriting and fixing software24Qwen3.8 Max Preview4 of 6 tests
- Images and chartsUnderstanding pictures, charts and video02Qwen3.8 Max Preview2 of 2 tests
- Following instructionsDoing exactly what it is asked02Qwen3.8 Max Preview2 of 2 tests
- FactsGetting facts right instead of making them up12Qwen3.8 Max Preview2 of 3 tests
- MathCompetition and research-level math30GPT-5.43 of 3 tests
- ReasoningHard problems that need careful thinking10GPT-5.41 of 4 tests · 3 ties
- Long documentsFinding answers in very long texts10GPT-5.41 of 1 test
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.
Qwen3.8 Max Preview costs 54% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Qwen3.8 Max Preview pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+62.9points ahead
- Real work tasks from 44 professionsGDPVal+21points ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+11.1points ahead
Where GPT-5.4 pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+19.2points ahead
- Research-level physics problemsCritPt+5.7points ahead
- Programming problems, refreshed regularlyLiveBench · Coding+4.6points ahead
Every test, side by side
All 37 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.8 Max Preview
- LMArena · WebDevQwen3.8 Max Preview by 208 rating points14641672+208 rating
- LiveBench · Agentic CodingQwen3.8 Max Preview by 10.953.864.7+10.9
- Terminal-Bench 2.1Qwen3.8 Max Preview by 10.578.388.8+10.5
- SWE-bench ProQwen3.8 Max Preview by 1057.767.7+10
- LiveBench · CodingGPT-5.4 by 4.677.572.9+4.6
- SciCodeGPT-5.4 by 4.556.652.1+4.5
AgentsQwen3.8 Max Preview
- GDPValQwen3.8 Max Preview by 2137.358.3+21
- OSWorld-VerifiedQwen3.8 Max Preview by 11.17586.1+11.1
- AA ApexAgentsQwen3.8 Max Preview by 9.133.342.4+9.1
- τ-Bench V3 · BankingQwen3.8 Max Preview by 8.239.647.8+8.2
- AA IT-Bench SREQwen3.8 Max Preview by 5.834.540.3+5.8
ReasoningGPT-5.4
- CritPtGPT-5.4 by 5.723.417.7+5.7
- GPQA Diamondtie9292.8tie
- Humanity's Last Examtie43.743.1tie
- LiveBench · Reasoningtie88.188.2tie
FactsQwen3.8 Max Preview
- AA-Omniscience · Non-hallucinationQwen3.8 Max Preview by 62.98.371.2+62.9
- AA-Omniscience · AccuracyGPT-5.4 by 19.250.931.7+19.2
- SimpleQA VerifiedQwen3.8 Max Preview by 2.245.147.3+2.2
Images and chartsQwen3.8 Max Preview
- MMMU-ProQwen3.8 Max Preview by 4.478.482.8+4.4
- LMArena · VisionQwen3.8 Max Preview by 11 rating points13031314+11 rating
MathGPT-5.4
- FrontierMath Tiers 1-3 (v2)GPT-5.4 by 3.978.674.7+3.9
- LiveBench · MathematicsGPT-5.4 by 2.994.291.3+2.9
- FrontierMath Tier 4GPT-5.4 by 2.74946.3+2.7
Following instructionsQwen3.8 Max Preview
- IFBenchQwen3.8 Max Preview by 8.973.982.8+8.9
- LiveBench · Instruction FollowingQwen3.8 Max Preview by 3.970.274.1+3.9
Other results11 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.
- ToolathlonQwen3.8 Max Preview by 17.954.672.5+17.9
- NL2RepoQwen3.8 Max Preview by 14.641.355.9+14.6
- CyberGymQwen3.8 Max Preview by 12.266.378.5+12.2
- AA IntelligenceQwen3.8 Max Preview by 6.43945.4+6.4
- AA-OmniscienceQwen3.8 Max Preview by 6.25.812+6.2
- AA Agentic IndexQwen3.8 Max Preview by 5.444.249.6+5.4
- DeepSWE 1.1Qwen3.8 Max Preview by 4.65256.6+4.6
- HLE (with tools)Qwen3.8 Max Preview by 4.152.156.2+4.1
- livebench_languageGPT-5.4 by 2.982.679.7+2.9
- livebench_data_analysistie79.378.4tie
- Artificial Analysis Coding Indextie71.171.8tie
Questions people ask
Which is better, GPT-5.4 or Qwen3.8 Max Preview?
Qwen3.8 Max Preview wins five of the eight areas we test: coding, agents, facts, images and charts and following instructions. GPT-5.4 wins reasoning, math and long documents.
Which is better for coding?
Qwen3.8 Max Preview. It wins 4 of the 6 coding tests both models report; GPT-5.4 wins 2.
Which is cheaper?
GPT-5.4 costs $2.50 per million input tokens and $15.00 per million output tokens; Qwen3.8 Max Preview costs $2.00 and $6.00. That makes Qwen3.8 Max Preview about 54% cheaper for the same work.
How do you compare the two?
We use the 37 benchmark tests both models have published scores on. The verdict counts the 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 11 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.