Compare / head-to-head
GLM-5.3vsGPT-5.5
GPT-5.5 leads 3 of 5 shared benchmarks. GLM-5.3 is 6x cheaper per token. Both offer a 1M-token context window.
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
2 – 3
GPT-5.5 leads
Cheaper per token
GLM-5.3
6x cheaper, input + output
Larger context
Tie
both 1M tokens
Provider uptime (30d)
— · 94.2446%
Z.ai · OpenAI
GLM-5.3
Z.ai · released 2026-08-14
text
Quality
Benchmark matrix
5 shared · 19 only GLM-5.3 · 16 only GPT-5.5| Benchmark | GLM-5.3 | GPT-5.5 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 5 | ||||
| GPQA Diamond | 90.9% ↗epoch | 90.7% ↗epoch | +0.2 pt | GLM-5.3 |
| FrontierMath Tier 4 v2 (Epoch AI run) | 29.3% ↗epoch | 72.5% ↗epoch | -43.2 pt | GPT-5.5 |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 68.8% ↗epoch | 85.3% ↗epoch | -16.5 pt | GPT-5.5 |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 91.1% ↗epoch | 84.4% ↗epoch | +6.7 pt | GLM-5.3 |
| SimpleQA Verified | 41% ↗epoch | 63% ↗epoch | -22 pt | GPT-5.5 |
| Only GLM-5.3 reports · 19 | ||||
| Agents' Last Exam | 28.5% ↗ | not reported | — | — |
| Agents' Last Exam (ALE-CLI) | 28.5% ↗ | not reported | — | — |
| AutomationBench v1.0.6 | 48.2% ↗ | not reported | — | — |
| CyberGym | 84.5% ↗ | not reported | — | — |
| DeepSWE v1.1 | 66.9% ↗ | not reported | — | — |
| ExploitBench | 54.4% ↗ | not reported | — | — |
| ExploitGym 2h | 105 ↗ | not reported | — | — |
| ExploitGym 6h | 130 ↗ | not reported | — | — |
| FrontierSWE | 78.1% ↗ | not reported | — | — |
| GDPval-AA v2 | 1769 ↗ | not reported | — | — |
| HLE (with tools) | 62.5% ↗ | not reported | — | — |
| NL2Repo | 58% ↗ | not reported | — | — |
| PostTrainBench | 39.8% ↗ | not reported | — | — |
| ProgramBench (Almost Solved) | 19% ↗ | not reported | — | — |
| SWE-Marathon v1.1 | 42.5% ↗ | not reported | — | — |
| Terminal-Bench 2.1 | 88.2% ↗ | not reported | — | — |
| Terminal-Bench 3.0 | 28.3% ↗ | not reported | — | — |
| Toolathlon Verified | 73% ↗ | not reported | — | — |
| Z.ai Code Bench (max effort) | 34.5% ↗ | not reported | — | — |
| Only GPT-5.5 reports · 16 | ||||
| AIME 2026 | not reported | 100% ↗matharena ⚠ | — | — |
| LMArena Elo | not reported | 1465.6 ↗ | — | — |
| ARC-AGI-1 (Verified) | not reported | 95% ↗ | — | — |
| ARC-AGI-2 (Verified) | not reported | 85% ↗ | — | — |
| BrowseComp | not reported | 84.4% ↗ | — | — |
| FrontierMath Tier 1-3 | not reported | 51.7% ↗ | — | — |
| FrontierMath Tier 4 | not reported | 35.4% ↗ | — | — |
| GDPval (wins or ties) | not reported | 84.9% ↗ | — | — |
| MCP Atlas | not reported | 75.3% ↗ | — | — |
| MMMU-Pro (no tools) | not reported | 81.2% ↗ | — | — |
| MMMU-Pro (with tools) | not reported | 83.2% ↗ | — | — |
| OSWorld-Verified | not reported | 78.7% ↗ | — | — |
| SWE-Bench Pro | not reported | 58.6% ↗ | — | — |
| Tau2-bench Telecom | not reported | 98% ↗ | — | — |
| Terminal-Bench 2.0 | not reported | 82.7% ↗ | — | — |
| Toolathlon | not reported | 55.6% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is GLM-5.3 minus GPT-5.5 in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | GLM-5.3 | GPT-5.5 | Edge |
|---|---|---|---|
| Context window | 1M tokens | 1M tokens | Tie |
| Max output | 131K tokens | 128K tokens | GLM-5.3 |
| Input price / 1M | $1.4 ↗ | $5 ↗ | GLM-5.3 |
| Output price / 1M | $4.4 ↗ | $30 ↗ | GLM-5.3 |
| Cached input / 1M | $0.26 ↗ | $0.5 ↗ | GLM-5.3 |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $5.8 | $35 | GLM-5.3 |
| Modalities | text | text · vision | GPT-5.5 |
| Released | 2026-08-14 | 2026-04-23 | — |
| Cited benchmark scores | 33 | 21 | — |
Reliability
Provider status
Last 60 days · refreshed daily on this page · live on /statusMore matchups
GLM-5.3 vs …
Models sharing the most benchmarksMore matchups
GPT-5.5 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run GLM-5.3 and GPT-5.5 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.