Reasoning
198-question Diamond subset; independent AA evaluation
- 01
- 02
- 03
- 03
- 03
- 14
GLM-5.3
Across the task benchmarks we track, GLM-5.3’s strongest relative results are in work automations, knowledge work, and reasoning. On AutomationBench-AA, it scores 62.2%, ranking #13 among 55 models with reported results in our dataset.
Updated: Data as of:
Strongest results
| Benchmark | Measures | Score |
|---|---|---|
| Agent Index ↗v1.2 Preview | Overall agent capability | 27.3 / 100#19 of 26 |
| AutomationBench-AA ↗1.0.6 / AA private 657 | Work automations | 62.2%#13 of 55 |
| GDPval-AA ↗2.1 | Knowledge work | 1,656.88 Elo#10 of 38 |
| AA-Briefcase ↗1.1 / combined Elo | Business deliverables | 1,504.57 Elo#15 of 57 |
Model details
| Model | GLM-5.3 |
|---|---|
| Agent Index v1.2 | 27.3 / 100 · Rank 19 · Preview |
| Creator | Z.ai (opens in a new tab) |
| Availability | Open source / open weights |
| Context size | 1,000,000 tokens (opens in a new tab) |
| Max output | 131,072 tokens (opens in a new tab) |
| Knowledge cutoff | Not reported |
| License | GLM-5.3 License · Model weights ↗ (opens in a new tab) |
Pricing and performance
Per provider and reported configuration. Prices are USD per 1M tokens.
| Provider / configuration | Contexttokens | InputUSD / 1M tokens | OutputUSD / 1M tokens | LatencyTTFT · seconds | Speedtokens / sec |
|---|---|---|---|---|---|
| GLM-5.3 (max) | 1,048,576 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 2.822 (opens in a new tab) | 106.426 (opens in a new tab) |
| GLM-5.3 (max) (FAST) | 1,000,000 | $2.10 (opens in a new tab) | $6.60 (opens in a new tab) | 0.795 (opens in a new tab) | 276.567 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $0.63 (opens in a new tab) | $1.98 (opens in a new tab) | 1.588 (opens in a new tab) | 156.031 (opens in a new tab) |
| GLM-5.3 (max) (FP4) | 786,432 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 0.723 (opens in a new tab) | 269.762 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 0.925 (opens in a new tab) | 154.356 (opens in a new tab) |
| GLM-5.3 (max) | 1,048,575 | $0.5625 (opens in a new tab) | $2.50 (opens in a new tab) | 1.01 (opens in a new tab) | 129.923 (opens in a new tab) |
| GLM-5.3 (max) | 1,048,576 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 0.837 (opens in a new tab) | 58.931 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $1.26 (opens in a new tab) | $3.96 (opens in a new tab) | 0.793 (opens in a new tab) | 139.209 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $0.70 (opens in a new tab) | $2.20 (opens in a new tab) | 2.735 (opens in a new tab) | 125.736 (opens in a new tab) |
| GLM-5.3 (max) | 163,840 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 1.539 (opens in a new tab) | 96.309 (opens in a new tab) |
| GLM-5.3 (max) (FAST) | 1,000,000 | $2.80 (opens in a new tab) | $8.80 (opens in a new tab) | 0.913 (opens in a new tab) | 482.734 (opens in a new tab) |
| GLM-5.3 (max) (FP4 ULTRA CHAT) | 1,048,576 | $2.10 (opens in a new tab) | $6.60 (opens in a new tab) | Not reported | Not reported |
| GLM-5.3 (max) (FP4) | 1,048,576 | $1.05 (opens in a new tab) | $3.30 (opens in a new tab) | 0.91 (opens in a new tab) | 373.189 (opens in a new tab) |
| GLM-5.3 (max) | 1,048,576 | $0.14 (opens in a new tab) | $4.40 (opens in a new tab) | 1.309 (opens in a new tab) | 121.54 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 0.682 (opens in a new tab) | 203.893 (opens in a new tab) |
| GLM-5.3 (max) | 1,048,576 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 0.634 (opens in a new tab) | 187.466 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $0.20 (opens in a new tab) | $0.64 (opens in a new tab) | 1.284 (opens in a new tab) | 214.631 (opens in a new tab) |
| GLM-5.3 (max) (FP4) | 1,000,000 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 1.492 (opens in a new tab) | 238.818 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 1.938 (opens in a new tab) | 90.783 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 0.873 (opens in a new tab) | 217.437 (opens in a new tab) |
| GLM-5.3 (max) | 262,144 | $2.09 (opens in a new tab) | $6.38 (opens in a new tab) | 1.268 (opens in a new tab) | 265.356 (opens in a new tab) |
| GLM-5.3 (max) (FP4) | 1,000,000 | $0.63 (opens in a new tab) | $1.98 (opens in a new tab) | 1.814 (opens in a new tab) | 207.47 (opens in a new tab) |
| GLM-5.3 (max) (FP4, FAST) | 1,000,000 | $2.80 (opens in a new tab) | $8.80 (opens in a new tab) | 1.386 (opens in a new tab) | 561.789 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $0.50 (opens in a new tab) | $1.00 (opens in a new tab) | 0.947 (opens in a new tab) | 688.217 (opens in a new tab) |
| GLM-5.3 (max) | 1,048,575 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 0.597 (opens in a new tab) | 188.682 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 0.891 (opens in a new tab) | 157.852 (opens in a new tab) |
| GLM-5.3 (max) | 1,000,000 | $1.40 (opens in a new tab) | $4.40 (opens in a new tab) | 1.477 (opens in a new tab) | 88.436 (opens in a new tab) |
Benchmark performance
GLM-5.3 in color, alongside leading models. Latest stored reports, best effort per report date.
GPQA Diamond ↗
Coding
113 tasks; mini-swe-agent
- 01
- 01
- 01
- 04
- 05
- 06
Terminal Use
Official 4.0 board; native agent harnesses
- 01
- 02
- 03
- 03
- 05
- 08
Work Automations
Private 657 tasks; 50 turns; objective completion
- 01
- 02
- 03
- 04
- 05
- 13
Knowledge Work
220 tasks; Stirrup agent; pairwise Elo
- 01
- 02
- 03
- 04
- 05
- 10
Reported benchmarks
All published results in our dataset, with versions, evaluation settings, and sources.
| Benchmark | Score | Evaluation setup | Source |
|---|---|---|---|
| AA-Briefcase1.1 / combined EloCombined Elo · Higher is better | 1,504.57 Elo | selection: best reported configuration per release · GLM-5.3 (Max) · tools: benchmark-defined | Artificial Analysis (opens in a new tab)independent |
| AutomationBench-AA1.0.6 / AA private 657Objectives completed without guardrail violations · Higher is better | 62.203% | GLM-5.3 (Max); Private 657 tasks; 50 turns; objective completion | Artificial Analysis (opens in a new tab)independent |
| AutomationBench-AA1.0.6 / AA private 657Objectives completed without guardrail violations · Higher is better | 51.979% | GLM-5.3 (Low); Private 657 tasks; 50 turns; objective completion | Artificial Analysis (opens in a new tab)independent |
| DeepSWEv1.1Pass@1 · Higher is better | 69% | mini-swe-agent; effort=max; ±3 percentage points | Datacurve ↗ (opens in a new tab)benchmark publisher |
| DeepSWEv1.1Pass@1 · Higher is better | 69% | max; ±3 percentage points | Datacurve ↗ (opens in a new tab)benchmark publisher |
| GDP.pdf100 tasks / All-pass / five attemptsAccuracy · Higher is better | 11.2% | selection: best reported configuration per release · GLM-5.3 (Max) · tools: benchmark-defined | Artificial Analysis (opens in a new tab)independent |
| GDPval-AA2.1Elo · Higher is better | 1,656.88 Elo | GLM-5.3 (Max); 220 tasks; Stirrup agent; pairwise Elo | Artificial Analysis (opens in a new tab)independent |
| GDPval-AA2.1Elo · Higher is better | 1,333.34 Elo | GLM-5.3 (Low); 220 tasks; Stirrup agent; pairwise Elo | Artificial Analysis (opens in a new tab)independent |
| GPQADiamond / Artificial AnalysisAccuracy · Higher is better | 91.717% | GLM-5.3 (Max); 198-question Diamond subset; independent AA evaluation | Artificial Analysis (opens in a new tab)independent |
| HLE-Diamond1,000 questions / no toolsAccuracy · Higher is better | 17.9% | selection: best reported configuration per release · GLM-5.3 (Max) · No tools | Artificial Analysis (opens in a new tab)independent |
| Humanity's Last ExamHLE — source does not specify subsetAccuracy · Higher is better | 62.5% | temperature=1.0; top_p=0.95; max generation 163840 tokens; max context 300000 tokens; context management; GPT-5.6 Luna medium judge. · subset: unspecified · Tools enabled | Z.ai ↗ (opens in a new tab)provider reported · 2026-08-14 |
| Humanity's Last ExamStable HLE — text-only subsetAccuracy · Higher is better | 42.3% | question count: 2158 · reported precision decimals: 1 · GLM-5.3 (max); original HLE; 2,158 text-only questions; no tools; pass@1; chart labels rounded to one decimal. · source model href: /models/glm-5-3 · subset: text · No tools | Artificial Analysis (opens in a new tab)independent |
| LiveBench: Agentic2.0 / 2026-10-14Accuracy · Higher is better | 35.9% | release: 2026-10-14 · Official LiveBench 2.0 row GLM-5.3. One run per suite; effort not explicitly labeled on this row. Historical release, not the October 25 index track. · source model id: glm-5.3 · suite: agentic | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-14 |
| LiveBench: Browser Use2.0 / 2026-10-14Accuracy · Higher is better | 0% | release: 2026-10-14 · Official LiveBench 2.0 row GLM-5.3. One run per suite; effort not explicitly labeled on this row. Historical release, not the October 25 index track. · source model id: glm-5.3 · suite: browser-use | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-14 |
| LiveBench: Coding2.0 / 2026-10-14Accuracy · Higher is better | 82.8% | release: 2026-10-14 · Official LiveBench 2.0 row GLM-5.3. One run per suite; effort not explicitly labeled on this row. Historical release, not the October 25 index track. · source model id: glm-5.3 · suite: coding | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-14 |
| Terminal-Bench2.1 / Artificial AnalysisPass@1 · Higher is better | 83.895% | GLM-5.3 (Max); 89 tasks; Terminus 2; e2b; three repeats | Artificial Analysis (opens in a new tab)independent |
| Terminal-Bench4.0Task success · Higher is better | 41.8% | Claude Code; effort=max; 41.8% ± 3.2% | Terminal-Bench ↗ (opens in a new tab)benchmark publisher |
| Terminal-Bench4.0 / mini-swe-agentTask success · Higher is better | 41.919% | harness: mini-swe-agent · repeats: 3 · selection: best reported configuration per release · GLM-5.3 (Max) | Artificial Analysis (opens in a new tab)independent |