GLM-5.1 vs GPT-5
Wins 3 of 7 areas
Coding · Agents · Following instructions
Wins 2 of 7 areas
Facts · Long documents
GLM-5.1 wins more areas, narrowly.GPT-5 is better at facts.
Scores updated · 34 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 software71GLM-5.17 of 8 tests
- AgentsCarrying out multi-step tasks on its own21GLM-5.12 of 3 tests
- Following instructionsDoing exactly what it is asked10GLM-5.11 of 1 test
- FactsGetting facts right instead of making them up12GPT-52 of 3 tests
- Long documentsFinding answers in very long texts01GPT-51 of 1 test
- ReasoningHard problems that need careful thinking22Even2 each
- MathCompetition and research-level math11Even1 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.
GLM-5.1 costs 49% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GLM-5.1 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+52.3points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+26.6points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+24.4points ahead
Where GPT-5 pulls ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+18.6points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+16.6points ahead
- Short factual questions, answered correctlySimpleQA Verified+16.1points ahead
Every test, side by side
All 34 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGLM-5.1
- Terminal-Bench 2.1GLM-5.1 by 26.661.835.2+26.6
- SWE-bench MultilingualGLM-5.1 by 1873.355.3+18
- SWE-bench ProGLM-5.1 by 16.658.441.8+16.6
- Terminal-Bench HardGLM-5.1 by 10.643.232.6+10.6
- LMArena · WebDevGLM-5.1 by 92 rating points15091417+92 rating
- SciCodeGLM-5.1 by 1.944.842.9+1.9
- LiveCodeBench v6GPT-5 by 1.385.787+1.3
- SWE-bench VerifiedGLM-5.1 by 1.376.274.9+1.3
AgentsGLM-5.1
- BrowseCompGLM-5.1 by 24.479.354.9+24.4
- GDPValGLM-5.1 by 103121+10
- τ-Bench V3 · BankingGPT-5 by 8.513.622.1+8.5
ReasoningEven
- Humanity's Last ExamGLM-5.1 by 1.630.128.5+1.6
- SimpleBenchGPT-5 by 1.655.156.7+1.6
- GPQA DiamondGLM-5.1 by 1.486.885.4+1.4
- CritPtGPT-5 by 1.14.65.7+1.1
FactsGPT-5
- AA-Omniscience · Non-hallucinationGLM-5.1 by 52.370.117.8+52.3
- AA-Omniscience · AccuracyGPT-5 by 16.623.740.3+16.6
- SimpleQA VerifiedGPT-5 by 16.13450.1+16.1
MathEven
- FrontierMath Tiers 1-3 (v2)GPT-5 by 18.636.855.4+18.6
- IMOAnswerBenchGLM-5.1 by 7.883.876+7.8
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.
- Terminal-Bench 2.0GLM-5.1 by 33.86935.2+33.8
- Artificial Analysis Coding IndexGLM-5.1 by 1855.837.8+18
- HLE (with tools)GLM-5.1 by 17.152.335.2+17.1
- τ²-Bench Telecom (AA run)GLM-5.1 by 12.997.784.8+12.9
- AA-OmniscienceGLM-5.1 by 9.60.8-8.7+9.6
- HMMT Nov. 2025GLM-5.1 by 4.89489.2+4.8
- AA IntelligenceGLM-5.1 by 3.126.123+3.1
- AA Agentic IndexGPT-5 by 1.325.226.5+1.3
- MMLU-ProGPT-5 by 1.18687.1+1.1
- HMMT 2025tie9493.3tie
- GPQA (unspecified)tie86.185.7tie
- frontiermath_tier_4_v1tie12.512.5tie
Questions people ask
Which is better, GLM-5.1 or GPT-5?
GLM-5.1 wins three of the seven areas where both have results: coding, agents and following instructions. GPT-5 wins facts and long documents. They are level on reasoning and math.
Which is better for coding?
GLM-5.1. It wins 7 of the 8 coding tests both models report; GPT-5 wins 1.
Which is cheaper?
GLM-5.1 costs $1.38 per million input tokens and $4.40 per million output tokens; GPT-5 costs $1.25 and $10.00. That makes GLM-5.1 about 49% cheaper for the same work.
How do you compare the two?
We use the 34 benchmark tests both models have published scores on. The verdict counts the 22 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.