ModelsCompareBest forBenchmarksStatusPricingAPI

GLM-5.3

Z.ai·Pending·Released 2026-08-14text
Context1,000k
Max output131.072k
Input / 1M$1.4 ↗
Output / 1M$4.4 ↗
Cached in$0.26
Benchmarks33

Benchmarks

Cited public sources · rank is exact-benchmark, apples-to-apples

FrontierMath Tier 4 v2 (Epoch AI run)#19/4429.3% ↗
FrontierMath Tiers 1-3 v2 (Epoch AI run)#17/5968.8% ↗
GPQA Diamond (Epoch AI run)#20/7790.9% ↗
OTIS Mock AIME 2024-2025 (Epoch AI run)#33/7691.1% ↗
SimpleQA Verified (Epoch AI run)#28/5341% ↗
Agents' Last Exam#1/228.5% ↗
CyberGym#1/284.5% ↗
DeepSWE v1.1#7/1366.9% ↗
ExploitBench#1/254.4% ↗
Terminal-Bench 3.0#1/328.3% ↗
Agents' Last Exam#1/228.5% ↗
Agents' Last Exam (ALE-CLI)#1/328.5% ↗
AutomationBench v1.0.6#2/348.2% ↗
CyberGym#1/284.5% ↗
CyberGym#1/284.5% ↗
DeepSWE v1.1#7/1366.9% ↗
DeepSWE v1.1#7/1366.9% ↗
ExploitBench#1/254.4% ↗
ExploitBench#1/254.4% ↗
ExploitGym 2h#1/2105 ↗
ExploitGym 6h#1/2130 ↗
FrontierSWE#2/378.1% ↗
GDPval-AA v2#4/71769 ↗
HLE (with tools)#1/462.5% ↗
NL2Repo#1/458% ↗
PostTrainBench#1/339.8% ↗
ProgramBench (Almost Solved)#1/219% ↗
SWE-Marathon v1.1#1/242.5% ↗
Terminal-Bench 2.1#3/1788.2% ↗
Terminal-Bench 3.0#1/328.3% ↗
Terminal-Bench 3.0#1/328.3% ↗
Toolathlon Verified#3/473% ↗
Z.ai Code Bench (max effort)34.5% ↗

Compare

vs GPT-6 Astravs GPT-5.6 Solvs GPT-5.6 Terravs GPT-5.6 Lunavs Claude Fable 5.1vs Claude Opus 5vs Claude Sonnet 5vs Claude Haiku 4.5

More from Z.ai

GLM-5.3-FlashGLM-5.2GLM-5GLM-4.7-FlashXGLM-5.1GLM-5-Turbo

Access GLM-5.3 and every other model through one endpoint with automatic failover: Respan gateway.