MiniMax M2 wins more areas, narrowly.
Scores updated · 33 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.
- Long documentsFinding answers in very long texts01MiniMax M21 of 1 test
- Following instructionsDoing exactly what it is asked01MiniMax M21 of 1 test
- CodingWriting and fixing software22Even2 each · 1 tie
- AgentsCarrying out multi-step tasks on its own11Even1 each
- ReasoningHard problems that need careful thinking00Even0 each · 3 ties
- FactsGetting facts right instead of making them up11Even1 each
Long documents and following instructions 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.
MiniMax M2 costs 46% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where MiniMax M2 pulls ahead
Where GLM-4.6 pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+3.7points ahead
- Code for real scientific research problemsSciCode+2.3points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+1.1points ahead
Every test, side by side
All 33 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingEven
- LMArena · WebDevGLM-4.6 by 44 rating points13411297+44 rating
- SWE-bench MultilingualMiniMax M2 by 2.753.856.5+2.7
- SciCodeGLM-4.6 by 2.338.436.1+2.3
- SWE-bench VerifiedMiniMax M2 by 1.46869.4+1.4
- Terminal-Bench Hardtie2525.8tie
ReasoningEven
- Humanity's Last Examtie14.513.7tie
- GPQA Diamondtie7877.7tie
- CritPttie1.10.9tie
FactsEven
- AA-Omniscience · AccuracyGLM-4.6 by 3.726.923.2+3.7
- AA-Omniscience · Non-hallucinationMiniMax M2 by 3.25.99.1+3.2
Following instructionsMiniMax M2
- IFBenchMiniMax M2 by 28.943.472.3+28.9
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.
- FinSearchComp-globalMiniMax M2 by 36.329.265.5+36.3
- AA Agentic IndexMiniMax M2 by 28.918.647.5+28.9
- Artificial Analysis Coding IndexGLM-4.6 by 16.645.829.2+16.6
- τ²-Bench Telecom (AA run)MiniMax M2 by 16.370.586.8+16.3
- AIME 2025GLM-4.6 by 15.993.978+15.9
- τ²-BenchMiniMax M2 by 11.875.287+11.8
- ArtifactsBenchMiniMax M2 by 759.866.8+7
- Multi-SWE-BenchMiniMax M2 by 6.23036.2+6.2
- Terminal-BenchMiniMax M2 by 5.840.546.3+5.8
- Terminal-Bench 2.0MiniMax M2 by 5.524.530+5.5
- AA-OmniscienceGLM-4.6 by 4.6-41.9-46.5+4.6
- GAIA (text only)MiniMax M2 by 3.871.975.7+3.8
- LiveCodeBenchMiniMax M2 by 3.579.583+3.5
- xbench-DeepSearchMiniMax M2 by 27072+2
- HLE (with tools)MiniMax M2 by 1.430.431.8+1.4
- MMLU-ProGLM-4.6 by 1.283.282+1.2
- BrowseComp-ZHGLM-4.6 by 149.548.5+1
- theagentcompanyMiniMax M2 by 13536+1
- AA Intelligencetie18.518.6tie
Questions people ask
Which is better, GLM-4.6 or MiniMax M2?
MiniMax M2 wins two of the six areas where both have results: long documents and following instructions. GLM-4.6 wins none. They are level on coding, agents, reasoning and facts.
Which is better for coding?
Neither. They win 2 coding tests each of the 5 both models report, and 1 is a tie.
Which is cheaper?
GLM-4.6 costs $0.57 per million input tokens and $2.20 per million output tokens; MiniMax M2 costs $0.30 and $1.20. That makes MiniMax M2 about 46% cheaper for the same work.
How do you compare the two?
We use the 33 benchmark tests both models have published scores on. The verdict counts the 14 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.