MiniMax M1 40K vs Qwen3 235B A22B
Wins 2 of 4 areas
Long documents · Following instructions
Wins 2 of 4 areas
Coding · Reasoning
The two are evenly matched.MiniMax M1 40K is better at long documents and following instructions; Qwen3 235B A22B at reasoning and coding.
Scores updated · 24 tests both models report · How we compare
Where each one wins
Tests won in each of the four 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.
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 three tests each model wins by the widest margin. Scores are out of 100.
Where MiniMax M1 40K pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+21.2points ahead
- Questions about very long textsLongBench v2+10.9points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+3.5points ahead
Where Qwen3 235B A22B pulls ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+3.8points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+3.2points ahead
- Code for real scientific research problemsSciCode+2.1points ahead
Every test, side by side
All 24 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 235B A22B
- SWE-bench VerifiedMiniMax M1 40K by 21.255.634.4+21.2
- Terminal-Bench HardQwen3 235B A22B by 3.82.36.1+3.8
- SciCodeQwen3 235B A22B by 2.137.839.9+2.1
ReasoningQwen3 235B A22B
- Humanity's Last ExamQwen3 235B A22B by 3.27.811+3.2
- GPQA DiamondQwen3 235B A22B by 1.868.270+1.8
Long documentsMiniMax M1 40K
- AA-LCRMiniMax M1 40K by 55.755.70+55.7
- LongBench v2MiniMax M1 40K by 10.96150.1+10.9
Following instructionsMiniMax M1 40K
- Multi-ChallengeMiniMax M1 40K by 3.544.741.2+3.5
- IFBenchMiniMax M1 40K by 2.541.238.7+2.5
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.
- OpenAI-MRCR (128k)MiniMax M1 40K by 48.476.127.7+48.4
- ZebraLogicMiniMax M1 40K by 42.480.137.7+42.4
- TAU-bench (airline)MiniMax M1 40K by 25.36034.7+25.3
- MMLU-ProMiniMax M1 40K by 12.480.668.2+12.4
- TAU-bench (retail)MiniMax M1 40K by 9.267.858.6+9.2
- LiveCodeBenchQwen3 235B A22B by 8.462.370.7+8.4
- τ²-Bench Telecom (AA run)MiniMax M1 40K by 7.631.624+7.6
- AIME 2025Qwen3 235B A22B by 6.974.681.5+6.9
- MATH-500 (EM)MiniMax M1 40K by 4.89691.2+4.8
- FullStackBenchMiniMax M1 40K by 4.767.662.9+4.7
- SimpleQAMiniMax M1 40K by 4.717.913.2+4.7
- LiveCodeBench (24/8~25/5)Qwen3 235B A22B by 3.662.365.9+3.6
- Artificial Analysis Coding IndexQwen3 235B A22B by 3.314.117.4+3.3
- AIME 2024Qwen3 235B A22B by 2.483.385.7+2.4
- AA Intelligencetie109.5tie
Questions people ask
Which is better, MiniMax M1 40K or Qwen3 235B A22B?
MiniMax M1 40K and Qwen3 235B A22B each win two of the four areas where both have results. MiniMax M1 40K wins long documents and following instructions; Qwen3 235B A22B wins coding and reasoning.
Which is better for coding?
Qwen3 235B A22B. It wins 2 of the 3 coding tests both models report; MiniMax M1 40K wins 1.
How do you compare the two?
We use the 24 benchmark tests both models have published scores on. The verdict counts the 9 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.