Gemini 3.1 Pro vs GLM-5.2
Wins 4 of 7 areas
Reasoning · Facts · Long documents · Following instructions
Wins 2 of 7 areas
Coding · Agents
Gemini 3.1 Pro wins more areas, narrowly.GLM-5.2 is cheaper and better at agents.
Scores updated · 58 tests both models report · How we compare
Where each one wins
Tests won in each of the seven 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 thinking51Gemini 3.1 Pro5 of 6 tests
- Following instructionsDoing exactly what it is asked20Gemini 3.1 Pro2 of 2 tests
- FactsGetting facts right instead of making them up21Gemini 3.1 Pro2 of 3 tests
- Long documentsFinding answers in very long texts10Gemini 3.1 Pro1 of 1 test
- AgentsCarrying out multi-step tasks on its own15GLM-5.25 of 6 tests
- CodingWriting and fixing software25GLM-5.25 of 7 tests
- MathCompetition and research-level math22Even2 each · 2 ties
Long documents 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.
GLM-5.2 costs 59% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Gemini 3.1 Pro pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+54.3points ahead
- Short factual questions, answered correctlySimpleQA Verified+39.3points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+30.6points ahead
Where GLM-5.2 pulls ahead
- Real work tasks from 44 professionsGDPVal+29points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+24.6points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+13.2points ahead
Every test, side by side
All 58 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGLM-5.2
- LMArena · WebDevGLM-5.2 by 156 rating points14471603+156 rating
- SWE-bench ProGLM-5.2 by 7.954.262.1+7.9
- LiveBench · Agentic CodingGLM-5.2 by 7.744.151.8+7.7
- SciCodeGemini 3.1 Pro by 7.558.751.2+7.5
- Terminal-Bench 2.1GLM-5.2 by 4.173.877.9+4.1
- LiveBench · CodingGLM-5.2 by 3.276.579.7+3.2
- Terminal-Bench HardGemini 3.1 Pro by 353.850.8+3
AgentsGLM-5.2
- GDPValGLM-5.2 by 2914.743.7+29
- τ-Bench V3 · BankingGLM-5.2 by 13.221.434.6+13.2
- AA IT-Bench SREGLM-5.2 by 12.430.342.7+12.4
- MCP AtlasGLM-5.2 by 7.669.276.8+7.6
- Terminal-Bench 4.0Gemini 3.1 Pro by 341+3
- AA ApexAgentsGLM-5.2 by 1.73233.7+1.7
ReasoningGemini 3.1 Pro
- ARC-AGI-2Gemini 3.1 Pro by 54.377.122.8+54.3
- SimpleBenchGemini 3.1 Pro by 20.879.658.8+20.8
- Humanity's Last ExamGemini 3.1 Pro by 5.94741.1+5.9
- LiveBench · ReasoningGemini 3.1 Pro by 5.48478.6+5.4
- GPQA DiamondGemini 3.1 Pro by 4.694.189.5+4.6
- CritPtGLM-5.2 by 3.217.720.9+3.2
FactsGemini 3.1 Pro
- SimpleQA VerifiedGemini 3.1 Pro by 39.373.534.2+39.3
- AA-Omniscience · AccuracyGemini 3.1 Pro by 30.654.924.3+30.6
- AA-Omniscience · Non-hallucinationGLM-5.2 by 24.649.173.7+24.6
MathEven
- IMOAnswerBenchGLM-5.2 by 108191+10
- FrontierMath Tier 4GLM-5.2 by 2.526.829.3+2.5
- HMMT Feb. 2026Gemini 3.1 Pro by 2.294.792.5+2.2
- LiveBench · MathematicsGemini 3.1 Pro by 1.29189.8+1.2
- AIME 2026tie98.399.2tie
- FrontierMath Tiers 1-3 (v2)tie59.659.2tie
Following instructionsGemini 3.1 Pro
- LiveBench · Instruction FollowingGemini 3.1 Pro by 16.879.162.3+16.8
- IFBenchGemini 3.1 Pro by 3.877.173.3+3.8
Other results27 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.
- GDPval-AA v2GLM-5.2 by 546 rating points9621508+546 rating
- GDPval-AA v2 EloGLM-5.2 by 545 rating points9651510+545 rating
- CyberGymGLM-5.2 by 38.438.877.2+38.4
- DeepSWEGLM-5.2 by 36.21046.2+36.2
- DeepSWE 1.1GLM-5.2 by 321244+32
- OfficeQA ProGemini 3.1 Pro by 31.172.541.4+31.1
- AA Agentic IndexGLM-5.2 by 29.110.339.4+29.1
- AA-OmniscienceGemini 3.1 Pro by 27.531.94.4+27.5
- Program BenchGLM-5.2 by 24.239.563.7+24.2
- ARC-AGI-1Gemini 3.1 Pro by 219877+21
- NL2RepoGLM-5.2 by 15.533.448.9+15.5
- PostTrainBenchGLM-5.2 by 12.721.634.3+12.7
- τ³-Bench BankingGLM-5.2 by 10.316.526.8+10.3
- livebench_languageGemini 3.1 Pro by 9.285.476.2+9.2
- SWE-MarathonGLM-5.2 by 9413+9
- HiL-Bench (Tools-allowed)GLM-5.2 by 8.435.343.7+8.4
- SWEBench Pro PublicGLM-5.2 by 7.954.262.1+7.9
- livebench_data_analysisGemini 3.1 Pro by 4.878.573.7+4.8
- AA IntelligenceGLM-5.2 by 429.733.7+4
- Global-MMLU-LiteGemini 3.1 Pro by 3.592.789.2+3.5
- τ²-Bench Telecom (AA run)GLM-5.2 by 3.595.699.1+3.5
- HLE (with tools)GLM-5.2 by 3.351.454.7+3.3
- Tool-Decathlontie48.848.2tie
- Toolathlontie48.848.2tie
- StrongREJECTtie9898.5tie
- HMMT Nov. 2025tie94.894.4tie
- Artificial Analysis Coding Indextie68.868.8tie
Questions people ask
Which is better, Gemini 3.1 Pro or GLM-5.2?
Gemini 3.1 Pro wins four of the seven areas where both have results: reasoning, facts, long documents and following instructions. GLM-5.2 wins coding and agents, and costs 59% less. They are level on math.
Which is better for coding?
GLM-5.2. It wins 5 of the 7 coding tests both models report; Gemini 3.1 Pro wins 2.
Which is cheaper?
Gemini 3.1 Pro costs $2.00 per million input tokens and $12.00 per million output tokens; GLM-5.2 costs $1.40 and $4.40. That makes GLM-5.2 about 59% cheaper for the same work.
How do you compare the two?
We use the 58 benchmark tests both models have published scores on. The verdict counts the 31 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 27 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.