GPT-5 vs Qwen3.5 35B A3B
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Long documents · Following instructions
Wins 0 of 8 areas
—
GPT-5 is the stronger all-rounder.Qwen3.5 35B A3B is cheaper.
Scores updated · 44 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 software53GPT-55 of 8 tests
- Images and chartsUnderstanding pictures, charts and video31GPT-53 of 4 tests
- ReasoningHard problems that need careful thinking20GPT-52 of 3 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own21GPT-52 of 3 tests
- FactsGetting facts right instead of making them up21GPT-52 of 3 tests
- Following instructionsDoing exactly what it is asked10GPT-51 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10GPT-51 of 1 test
- MathCompetition and research-level math00Even0 each · 1 tie
Math and long documents 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.
Qwen3.5 35B A3B 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 pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+20.2points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+17.2points ahead
- Real work tasks from 44 professionsGDPVal+15.4points ahead
Where Qwen3.5 35B A3B pulls ahead
- Reads and reasons about text in imagesOCRBenchv2+9.8points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+6.1points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+5.6points ahead
Every test, side by side
All 44 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5
- LMArena · WebDevGPT-5 by 168 rating points14181250+168 rating
- LiveCodeBench v6GPT-5 by 12.48774.6+12.4
- Terminal-Bench HardGPT-5 by 6.132.626.5+6.1
- SWE-bench VerifiedGPT-5 by 5.774.969.2+5.7
- Terminal-Bench 2.1Qwen3.5 35B A3B by 5.635.240.8+5.6
- SciCodeGPT-5 by 5.242.937.7+5.2
- SWE-bench MultilingualQwen3.5 35B A3B by 555.360.3+5
- SWE-bench ProQwen3.5 35B A3B by 2.841.844.6+2.8
AgentsGPT-5
- τ-Bench V3 · BankingGPT-5 by 17.222.14.9+17.2
- GDPValGPT-5 by 15.4215.6+15.4
- BrowseCompQwen3.5 35B A3B by 6.154.961+6.1
ReasoningGPT-5
- Humanity's Last ExamGPT-5 by 7.528.521+7.5
- CritPtGPT-5 by 4.85.70.9+4.8
- GPQA Diamondtie85.484.5tie
FactsGPT-5
- AA-Omniscience · AccuracyGPT-5 by 20.240.320.1+20.2
- Vectara HHEM hallucination ratelower is betterQwen3.5 35B A3B by 4.615.110.5+4.6
- AA-Omniscience · Non-hallucinationGPT-5 by 3.217.814.6+3.2
Images and chartsGPT-5
- OCRBenchv2Qwen3.5 35B A3B by 9.855.565.3+9.8
- CharXiv (RQ)GPT-5 by 3.681.177.5+3.6
- MMMUGPT-5 by 2.884.281.4+2.8
- MMMU-ProGPT-5 by 1.574.272.7+1.5
Following instructionsGPT-5
- Multi-ChallengeGPT-5 by 9.669.660+9.6
- IFBenchtie73.172.5tie
Other results19 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-OmniscienceGPT-5 by 39.4-8.7-48.1+39.4
- HumanEvalGPT-5 by 26.993.466.5+26.9
- AA Agentic IndexGPT-5 by 14.726.511.8+14.7
- vectara_avg_summary_lengthGPT-5 by 67.8 rating points162.794.9+67.8 rating
- BrowseComp-ZHQwen3.5 35B A3B by 6.56369.5+6.5
- Terminal-Bench 2.0Qwen3.5 35B A3B by 5.335.240.5+5.3
- vectara_factual_consistencyQwen3.5 35B A3B by 4.684.989.5+4.6
- τ²-Bench Telecom (AA run)Qwen3.5 35B A3B by 4.484.889.2+4.4
- VideoMMMUGPT-5 by 4.284.680.4+4.2
- HMMT 2025GPT-5 by 4.193.389.2+4.1
- AA IntelligenceGPT-5 by 3.72319.3+3.7
- MMLU-ReduxGPT-5 by 295.393.3+2
- MMLU-ProGPT-5 by 1.887.185.3+1.8
- GPQA (unspecified)GPT-5 by 1.585.784.2+1.5
- ERQAtie65.764.8tie
- Artificial Analysis Coding Indextie37.837tie
- HMMT Feb. 2025tie88.389tie
- vectara_answer_ratetie99.999.8tie
- HMMT Nov. 2025tie89.289.2tie
Questions people ask
Which is better, GPT-5 or Qwen3.5 35B A3B?
GPT-5 wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, long documents and following instructions. Qwen3.5 35B A3B wins none, but costs 80% less. They are level on math.
Which is better for coding?
GPT-5. It wins 5 of the 8 coding tests both models report; Qwen3.5 35B A3B wins 3.
Which is cheaper?
GPT-5 costs $1.25 per million input tokens and $10.00 per million output tokens; Qwen3.5 35B A3B costs $0.25 and $2.00. That makes Qwen3.5 35B A3B about 80% cheaper for the same work.
How do you compare the two?
We use the 44 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 19 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.