GLM-5.2 vs Qwen3.6 27B
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents · Following instructions
Wins 0 of 7 areas
—
GLM-5.2 is the stronger all-rounder.Qwen3.6 27B is cheaper.
Scores updated · 35 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.
- CodingWriting and fixing software60GLM-5.26 of 6 tests
- MathCompetition and research-level math50GLM-5.25 of 5 tests
- ReasoningHard problems that need careful thinking40GLM-5.24 of 4 tests
- AgentsCarrying out multi-step tasks on its own30GLM-5.23 of 3 tests
- FactsGetting facts right instead of making them up20GLM-5.22 of 2 tests
- Following instructionsDoing exactly what it is asked20GLM-5.22 of 2 tests
- Long documentsFinding answers in very long texts10GLM-5.21 of 1 test
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.
Qwen3.6 27B costs 28% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where GLM-5.2 pulls ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+24.1points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+23points ahead
- Research-level physics problemsCritPt+19.8points ahead
Where Qwen3.6 27B pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 35 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGLM-5.2
- Terminal-Bench 2.1GLM-5.2 by 17.277.960.7+17.2
- Terminal-Bench HardGLM-5.2 by 1650.834.8+16
- LiveBench · Agentic CodingGLM-5.2 by 12.551.839.3+12.5
- SWE-bench ProGLM-5.2 by 8.662.153.5+8.6
- SciCodeGLM-5.2 by 8.451.242.8+8.4
- LiveBench · CodingGLM-5.2 by 7.979.771.8+7.9
AgentsGLM-5.2
- GDPValGLM-5.2 by 19.343.724.4+19.3
- τ-Bench V3 · BankingGLM-5.2 by 17.934.616.7+17.9
- Terminal-Bench 4.0GLM-5.2 by 110+1
ReasoningGLM-5.2
- CritPtGLM-5.2 by 19.820.91.1+19.8
- Humanity's Last ExamGLM-5.2 by 1841.123.1+18
- LiveBench · ReasoningGLM-5.2 by 8.378.670.3+8.3
- GPQA DiamondGLM-5.2 by 5.389.584.2+5.3
FactsGLM-5.2
- AA-Omniscience · Non-hallucinationGLM-5.2 by 2373.750.7+23
- AA-Omniscience · AccuracyGLM-5.2 by 4.724.319.6+4.7
MathGLM-5.2
- FrontierMath Tiers 1-3 (v2)GLM-5.2 by 24.159.235.1+24.1
- IMOAnswerBenchGLM-5.2 by 10.29180.8+10.2
- LiveBench · MathematicsGLM-5.2 by 9.989.879.9+9.9
- HMMT Feb. 2026GLM-5.2 by 8.292.584.3+8.2
- AIME 2026GLM-5.2 by 5.199.294.1+5.1
Following instructionsGLM-5.2
- LiveBench · Instruction FollowingGLM-5.2 by 9.162.353.2+9.1
- IFBenchGLM-5.2 by 5.773.367.6+5.7
Other results12 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.
- DeepSWE 1.1GLM-5.2 by 30.74413.3+30.7
- AA-OmniscienceGLM-5.2 by 24.44.4-20+24.4
- JobBenchGLM-5.2 by 21.643.421.8+21.6
- AA Agentic IndexGLM-5.2 by 19.339.420.1+19.3
- Artificial Analysis Coding IndexGLM-5.2 by 15.168.853.7+15.1
- livebench_languageGLM-5.2 by 12.976.263.3+12.9
- NL2RepoGLM-5.2 by 12.748.936.2+12.7
- AA IntelligenceGLM-5.2 by 12.333.721.4+12.3
- τ²-Bench Telecom (AA run)GLM-5.2 by 4.999.194.2+4.9
- HMMT 2025GLM-5.2 by 3.794.490.7+3.7
- HMMT Nov. 2025GLM-5.2 by 3.794.490.7+3.7
- livebench_data_analysisGLM-5.2 by 3.373.770.4+3.3
Questions people ask
Which is better, GLM-5.2 or Qwen3.6 27B?
GLM-5.2 wins all seven areas where both have results: coding, agents, reasoning, facts, math, long documents and following instructions. Qwen3.6 27B wins none, but costs 28% less.
Which is better for coding?
GLM-5.2. It wins 6 of the 6 coding tests both models report; Qwen3.6 27B wins none.
Which is cheaper?
GLM-5.2 costs $1.40 per million input tokens and $4.40 per million output tokens; Qwen3.6 27B costs $0.60 and $3.60. That makes Qwen3.6 27B about 28% cheaper for the same work.
How do you compare the two?
We use the 35 benchmark tests both models have published scores on. The verdict counts the 23 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 12 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.