GPT-5.2 wins more areas, narrowly.GLM-5.1 is cheaper and better at agents.
Scores updated · 38 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.
- MathCompetition and research-level math04GPT-5.24 of 4 tests
- ReasoningHard problems that need careful thinking13GPT-5.23 of 4 tests
- FactsGetting facts right instead of making them up12GPT-5.22 of 3 tests
- Long documentsFinding answers in very long texts01GPT-5.21 of 1 test
- AgentsCarrying out multi-step tasks on its own31GLM-5.13 of 4 tests
- CodingWriting and fixing software33Even3 each
- Following instructionsDoing exactly what it is asked00Even0 each · 1 tie
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.
GLM-5.1 costs 63% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.2 pulls ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+30.6points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+20.6points ahead
- Real work tasks from 44 professionsGDPVal+17.3points ahead
Where GLM-5.1 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+51.3points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+28.5points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+13.5points ahead
Every test, side by side
All 38 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingEven
- SWE-bench ProGLM-5.1 by 28.558.429.9+28.5
- LMArena · WebDevGLM-5.1 by 94 rating points15091415+94 rating
- SciCodeGPT-5.2 by 7.344.852.1+7.3
- SWE-bench MultilingualGLM-5.1 by 6.673.366.7+6.6
- SWE-bench VerifiedGPT-5.2 by 3.876.280+3.8
- Terminal-Bench HardGPT-5.2 by 3.843.247+3.8
AgentsGLM-5.1
- GDPValGPT-5.2 by 17.33148.3+17.3
- BrowseCompGLM-5.1 by 13.579.365.8+13.5
- MCP AtlasGLM-5.1 by 11.271.860.6+11.2
- τ-Bench V3 · BankingGLM-5.1 by 2.513.611.1+2.5
ReasoningGPT-5.2
- SimpleBenchGLM-5.1 by 9.355.145.8+9.3
- Humanity's Last ExamGPT-5.2 by 7.630.137.7+7.6
- CritPtGPT-5.2 by 74.611.6+7
- GPQA DiamondGPT-5.2 by 3.586.890.3+3.5
FactsGPT-5.2
- AA-Omniscience · Non-hallucinationGLM-5.1 by 51.370.118.8+51.3
- AA-Omniscience · AccuracyGPT-5.2 by 20.623.744.3+20.6
- SimpleQA VerifiedGPT-5.2 by 3.13437.1+3.1
MathGPT-5.2
- FrontierMath Tiers 1-3 (v2)GPT-5.2 by 30.636.867.4+30.6
- HMMT Feb. 2026GPT-5.2 by 14.482.697+14.4
- AIME 2026GPT-5.2 by 395.398.3+3
- IMOAnswerBenchGPT-5.2 by 2.583.886.3+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.
- AA Agentic IndexGPT-5.2 by 3525.260.2+35
- Terminal-Bench 2.0GLM-5.1 by 156954+15
- τ²-Bench Telecom (AA run)GLM-5.1 by 12.997.784.8+12.9
- GDPval-AA (Elo)GLM-5.1 by 73 rating points15351462+73 rating
- Artificial Analysis Coding IndexGLM-5.1 by 7.155.848.7+7.1
- HLE (with tools)GLM-5.1 by 6.852.345.5+6.8
- frontiermath_tier_4_v1GPT-5.2 by 6.312.518.8+6.3
- GPQA (unspecified)GPT-5.2 by 6.386.192.4+6.3
- ToolathlonGPT-5.2 by 5.640.746.3+5.6
- HMMT 2025GPT-5.2 by 5.49499.4+5.4
- LiveBenchGPT-5.2 by 4.670.274.8+4.6
- AA IntelligenceGPT-5.2 by 4.326.130.4+4.3
- AA-OmniscienceGLM-5.1 by 1.80.8-0.9+1.8
- HMMT Nov. 2025GPT-5.2 by 1.89495.8+1.8
- MMLU-Protie8686.7tie
Questions people ask
Which is better, GLM-5.1 or GPT-5.2?
GPT-5.2 wins four of the seven areas where both have results: reasoning, facts, math and long documents. GLM-5.1 wins agents, and costs 63% less. They are level on coding and following instructions.
Which is better for coding?
Neither. They win 3 coding tests each of the 6 both models report.
Which is cheaper?
GLM-5.1 costs $1.38 per million input tokens and $4.40 per million output tokens; GPT-5.2 costs $1.75 and $14.00. That makes GLM-5.1 about 63% cheaper for the same work.
How do you compare the two?
We use the 38 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 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.