Zhipu AI (Z.ai)open-weight
Released: 2026-08-18Artificial Analysis

GLM 5.3

OpenRouter ranked #11 foundation model processing 3.71T tokens weekly. Deployed by z-ai.

Intelligence Index
31.7/ 100
Artificial Analysis Intelligence Index (independent composite)
Throughput Speed
84tok/sec
Output streaming throughput
Time to First Answer
63s
Time to first answer token, including any thinking
Cost to Run the Index
5.01USD
Artificial Analysis cost of the full Index run

Measured results · Artificial Analysis

Independent evaluations of GLM 5.3 (Max configuration). Values are reported as published; nothing here is estimated or derived. Snapshot 2026-11-15.

Intelligence Index
31.6
Cost to run the Index
$5.01
Time per task
23.6 min
Omniscience index
6.0
Reasoning model
Yes
Cached-input price
$0.26 / 1M

Agentic work

Multimodal & science

  • AA-3D (Elo)794 Elo
  • Code2Video-AA (Elo)1034 Elo

Reasoning-effort variants evaluated

VariantIntelligence IndexCost to run IndexOutput speed (tok/s)
Max31.7$5.0184.2
Low23.1$1.5768.0
Full analysis on Artificial Analysis

Headline Evaluation Breakdown

Six independently measured evaluations (Artificial Analysis)
Reasoning · CritPt v2 (Hard)33.5%
Coding & agents · Terminal-Bench 4.041.9%
Long context · AA-LCR79.7%
Knowledge · Humanity's Last Exam42.3%

Sourced Benchmark Evaluations (24)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
AA-Omniscience (Accuracy)Active32.3%
93
2026-11-15independentSource ↗
ITBench-AAActive46.1%
93
2026-11-15independentSource ↗
EnterpriseOps-Gym-AAActive36.4%
93
2026-11-15independentSource ↗
Harvey LAB-AAActive0.3%
93
2026-11-15independentSource ↗
AA-AnalystAgentActive38.8%
93
2026-11-15independentSource ↗
AutomationBench-AAActive62.2%
93
2026-11-15independentSource ↗
MLCR-AA (Medical Long Context Reasoning)Active17.2%
93
2026-11-15independentSource ↗
GDP.pdfActive11.2%
93
2026-11-15independentSource ↗
FrontierCode-AA 2.0Active21.6%
93
2026-11-15independentSource ↗
HealthBench ProfessionalActive52.8%
93
2026-11-15independentSource ↗
HealthBenchActive54.5%
93
2026-11-15independentSource ↗
AA-Omniscience (Non-Hallucination)Active61.1%
93
2026-11-15independentSource ↗
FrontierCodeActive40.1%
98
2026-11-15independentSource ↗
HLE-DiamondActive17.9%
93
2026-11-15independentSource ↗
CritPt v2 (Hard)Active33.5%
93
2026-11-15independentSource ↗
CritPtActive19.1%
93
2026-11-15independentSource ↗
AA-LCR (Long Context Reasoning)Active79.7%
93
2026-11-15independentSource ↗
Terminal-Bench-Science 0.1Active9.5%
93
2026-11-15independentSource ↗
Terminal-Bench 4.0Active41.9%
93
2026-11-15independentSource ↗
Terminal-Bench 2.1Active83.9%
93
2026-11-15independentSource ↗
τ³-BankingActive50.3%
93
2026-11-15independentSource ↗
SciCodeActive59%
93
2026-11-15independentSource ↗
Humanity's Last ExamActive42.3%
93
2026-11-15independentSource ↗
GPQANearing Saturation91.7%
78
2026-11-15independentSource ↗
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run glm-5-3

# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5-3",
    "messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
  }'

Evaluation & Sourcing Notes

Scores listed for GLM 5.3 represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.