Qwen3 235B A22B vs Qwen3.5 0.8B
Wins 4 of 6 areas
Coding · Agents · Reasoning · Following instructions
Wins 0 of 6 areas
—
Qwen3 235B A22B is the stronger all-rounder.
Scores updated · 30 tests both models report · How we compare
Where each one wins
Tests won in each of the six 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.
- ReasoningHard problems that need careful thinking20Qwen3 235B A22B2 of 3 tests · 1 tie
- CodingWriting and fixing software20Qwen3 235B A22B2 of 2 tests
- Following instructionsDoing exactly what it is asked20Qwen3 235B A22B2 of 2 tests
- AgentsCarrying out multi-step tasks on its own10Qwen3 235B A22B1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
- Long documentsFinding answers in very long texts11Even1 each
Agents 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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Qwen3 235B A22B pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+58.9points ahead
- Questions about very long textsLongBench v2+24points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+22.3points ahead
Where Qwen3.5 0.8B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+16.3points ahead
Every test, side by side
All 30 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 235B A22B
- SciCodeQwen3 235B A22B by 39.939.90+39.9
- Terminal-Bench HardQwen3 235B A22B by 6.16.10+6.1
ReasoningQwen3 235B A22B
- GPQA DiamondQwen3 235B A22B by 58.97011.1+58.9
- Humanity's Last ExamQwen3 235B A22B by 9.9111.1+9.9
- CritPttie00tie
FactsEven
- AA-Omniscience · Non-hallucinationQwen3.5 0.8B by 16.322.438.7+16.3
- AA-Omniscience · AccuracyQwen3 235B A22B by 14.318.54.2+14.3
Long documentsEven
- LongBench v2Qwen3 235B A22B by 2450.126.1+24
- AA-LCRQwen3.5 0.8B by 909+9
Following instructionsQwen3 235B A22B
- Multi-ChallengeQwen3 235B A22B by 22.341.218.9+22.3
- IFBenchQwen3 235B A22B by 17.238.721.5+17.2
Other results18 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.
- AIME 2024Qwen3 235B A22B by 8085.75.7+80
- AIME 2025Qwen3 235B A22B by 72.881.58.7+72.8
- τ²-Bench (Retail)Qwen3 235B A22B by 50577+50
- BFCLv4Qwen3 235B A22B by 45.570.825.3+45.5
- MMMLUQwen3 235B A22B by 42.486.744.3+42.4
- MATH-500 (EM)Qwen3 235B A22B by 33.691.257.6+33.6
- IncludeQwen3 235B A22B by 32.973.540.6+32.9
- MMLU-ProXQwen3 235B A22B by 32.166.734.6+32.1
- MMLU-ReduxQwen3 235B A22B by 27.987.459.5+27.9
- MMLU-ProQwen3 235B A22B by 25.968.242.3+25.9
- τ²-Bench Telecom (AA run)Qwen3.5 0.8B by 23.72447.7+23.7
- SuperGPQAQwen3 235B A22B by 22.844.121.3+22.8
- Artificial Analysis Coding IndexQwen3 235B A22B by 17.417.40+17.4
- AA Agentic IndexQwen3 235B A22B by 12.718.45.7+12.7
- IFEvalQwen3 235B A22B by 1283.271.2+12
- AA-OmniscienceQwen3 235B A22B by 9.8-44.7-54.5+9.8
- τ²-BenchQwen3 235B A22B by 7.822.114.3+7.8
- AA IntelligenceQwen3 235B A22B by 3.49.56.1+3.4
Questions people ask
Which is better, Qwen3 235B A22B or Qwen3.5 0.8B?
Qwen3 235B A22B wins four of the six areas where both have results: coding, agents, reasoning and following instructions. Qwen3.5 0.8B wins none. They are level on facts and long documents.
Which is better for coding?
Qwen3 235B A22B. It wins 2 of the 2 coding tests both models report; Qwen3.5 0.8B wins none.
How do you compare the two?
We use the 30 benchmark tests both models have published scores on. The verdict counts the 12 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 18 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.