GPT-5.2 vs Qwen3.5 35B A3B
Wins 6 of 8 areas
Coding · Agents · Reasoning · Facts · Math · Following instructions
Wins 1 of 8 areas
Images and charts
GPT-5.2 is the stronger all-rounder.Qwen3.5 35B A3B is cheaper and better at images and charts.
Scores updated · 49 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 software51GPT-5.25 of 6 tests
- ReasoningHard problems that need careful thinking30GPT-5.23 of 3 tests
- MathCompetition and research-level math30GPT-5.23 of 3 tests
- AgentsCarrying out multi-step tasks on its own31GPT-5.23 of 4 tests
- FactsGetting facts right instead of making them up20GPT-5.22 of 3 tests · 1 tie
- Following instructionsDoing exactly what it is asked10GPT-5.21 of 1 test
- Images and chartsUnderstanding pictures, charts and video23Qwen3.5 35B A3B3 of 5 tests
- Long documentsFinding answers in very long texts11Even1 each
Following instructions 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.
Qwen3.5 35B A3B costs 86% 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-5.2 pulls ahead
- Real work tasks from 44 professionsGDPVal+42.7points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+24.2points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+20.5points ahead
Where Qwen3.5 35B A3B pulls ahead
- Reads and reasons about text in imagesOCRBenchv2+14.8points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+14.7points ahead
- Reads text in imagesOCRBench+10.3points ahead
Every test, side by side
All 49 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.2
- Terminal-Bench HardGPT-5.2 by 20.54726.5+20.5
- LMArena · WebDevGPT-5.2 by 165 rating points14151250+165 rating
- SWE-bench ProQwen3.5 35B A3B by 14.729.944.6+14.7
- SciCodeGPT-5.2 by 14.452.137.7+14.4
- SWE-bench VerifiedGPT-5.2 by 10.88069.2+10.8
- SWE-bench MultilingualGPT-5.2 by 6.466.760.3+6.4
AgentsGPT-5.2
- GDPValGPT-5.2 by 42.748.35.6+42.7
- τ-Bench V3 · BankingGPT-5.2 by 6.211.14.9+6.2
- BrowseCompGPT-5.2 by 4.865.861+4.8
- MCP AtlasQwen3.5 35B A3B by 1.860.662.4+1.8
ReasoningGPT-5.2
- Humanity's Last ExamGPT-5.2 by 16.737.721+16.7
- CritPtGPT-5.2 by 10.711.60.9+10.7
- GPQA DiamondGPT-5.2 by 5.890.384.5+5.8
FactsGPT-5.2
- AA-Omniscience · AccuracyGPT-5.2 by 24.244.320.1+24.2
- AA-Omniscience · Non-hallucinationGPT-5.2 by 4.218.814.6+4.2
- Vectara HHEM hallucination ratelower is bettertie10.810.5tie
Images and chartsQwen3.5 35B A3B
- OCRBenchv2Qwen3.5 35B A3B by 14.850.565.3+14.8
- OCRBenchQwen3.5 35B A3B by 10.380.791+10.3
- CharXiv (RQ)GPT-5.2 by 4.682.177.5+4.6
- MathVistaQwen3.5 35B A3B by 3.482.886.2+3.4
- MMMU-ProGPT-5.2 by 1.974.672.7+1.9
MathGPT-5.2
- HMMT Feb. 2026GPT-5.2 by 15.29781.8+15.2
- IMOAnswerBenchGPT-5.2 by 9.586.376.8+9.5
- AIME 2026GPT-5.2 by 598.393.3+5
Long documentsEven
- AA-LCRGPT-5.2 by 10.782.772+10.7
- LongBench v2Qwen3.5 35B A3B by 4.554.559+4.5
Other results22 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 Agentic IndexGPT-5.2 by 48.460.211.8+48.4
- AA-OmniscienceGPT-5.2 by 47.2-0.9-48.1+47.2
- ScreenSpot-Pro (No tools)GPT-5.2 by 17.786.368.6+17.7
- Terminal-Bench 2.0GPT-5.2 by 13.55440.5+13.5
- Artificial Analysis Coding IndexGPT-5.2 by 11.748.737+11.7
- AA IntelligenceGPT-5.2 by 11.130.419.3+11.1
- HMMT Feb. 2025GPT-5.2 by 1110089+11
- HMMT 2025GPT-5.2 by 10.299.489.2+10.2
- vectara_avg_summary_lengthGPT-5.2 by 91.4 rating points186.394.9+91.4 rating
- MMVUGPT-5.2 by 8.580.872.3+8.5
- GPQA (unspecified)GPT-5.2 by 8.292.484.2+8.2
- HMMT Nov. 2025GPT-5.2 by 6.695.889.2+6.6
- VideoMMMUGPT-5.2 by 5.585.980.4+5.5
- MMMLUGPT-5.2 by 4.489.685.2+4.4
- τ²-Bench Telecom (AA run)Qwen3.5 35B A3B by 4.484.889.2+4.4
- OmniDocBench 1.5Qwen3.5 35B A3B by 3.685.789.3+3.6
- SimpleVQAQwen3.5 35B A3B by 2.555.858.3+2.5
- MMLU-ProGPT-5.2 by 1.486.785.3+1.4
- ZeroBenchGPT-5.2 by 198+1
- MATH-Visiontie8383.9tie
- vectara_factual_consistencytie89.289.5tie
- vectara_answer_ratetie10099.8tie
Questions people ask
Which is better, GPT-5.2 or Qwen3.5 35B A3B?
GPT-5.2 wins six of the eight areas we test: coding, agents, reasoning, facts, math and following instructions. Qwen3.5 35B A3B wins images and charts, and costs 86% less. They are level on long documents.
Which is better for coding?
GPT-5.2. It wins 5 of the 6 coding tests both models report; Qwen3.5 35B A3B wins 1.
Which is cheaper?
GPT-5.2 costs $1.75 per million input tokens and $14.00 per million output tokens; Qwen3.5 35B A3B costs $0.25 and $2.00. That makes Qwen3.5 35B A3B about 86% cheaper for the same work.
How do you compare the two?
We use the 49 benchmark tests both models have published scores on. The verdict counts the 27 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 22 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.