0

GLM 5

Open weights

GLM-5 is an AI model from Zai, released with open weights.

Publisher
Zai
Family
Glm
Context
203K
Input
Output
License
mit
Released
11 Feb 2026

Cite

Notes

Only stored in your browser.

Attribution

Benchmark scores
frontiermathAALiveBenchSWE-benchmatharenaprime-hubOpenRewardVals AI
Attribution policy →
Context
203K
Input
$1
/1M tok
Output
$3.20
/1M tok
Blended
$1.55
#334/ 414/1M · 3:1
Speed
67
#147/ 414tok/s
Cache read
$0.12
0.12x input

Price & speed via Artificial Analysis · ranked across tracked models · caching via OpenRouter

Reported on 27 evals across 9 domains - best 97.4% on τ²-bench (Tau²-bench) · #6 of 320

Reported eval scores

27
Flashinfer Bench Env24.7
Infraresolutionbench82.7
HMMT97.9
KellyBench48395
Terminal Bench 2 Verified61.1
SWE Bench Multilingual73.3
τ²-bench (Tau²-bench)97.4
Terminal-Bench (Hard)39.4
LiveBench - Language77.5
LiveBench - Coding73.6
IFBench55.2
SWE-bench Multilingual69.7
LiveBench - Data Analysis67.9
LiveBench - Math83.5
LiveBench - Reasoning69.1
LiveBench - Instruction Following55.3
LiveBench71.2
Humanity's Last Exam (HLE)7.2
SWE-bench Verified72.8
SciCode38.3
GPQA Diamond66.6
CorpFin v262.9
TaxEval v270
Vibe Code Bench v1.123.4
FrontierMath16.4
MathArena65.2