z-ai
Z.ai: GLM 5
z-ai/glm-5
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
GLM-5 (z-ai/glm-5) is Z.ai's flagship open-weights mixture-of-experts model with 744B parameters and 40B active per token, released under an MIT license and listed on OpenRouter as z-ai/glm-5-20260211. The vendor positions it for complex systems engineering and long-horizon agentic work, reporting strong benchmark results in reasoning, coding and tool use; OpenRouter's Design Arena also lists measured head-to-head results across coding and UI categories. The catalog reports a 204,800-token context window, text-only input and output, and an optional reasoning mode. Weights can be served via vLLM, SGLang, KTransformers, Transformers or xLLM.[1][2]
Strengths
- Open weights under an MIT license, with documented local serving through vLLM, SGLang, KTransformers, Transformers and xLLM.[2]
- Catalog-reported 204,800-token context window; the vendor says its DeepSeek Sparse Attention integration lowers deployment cost while preserving long-context capacity.[1][2]
- OpenRouter reports support for tool calling, structured outputs and an optional reasoning mode; vendor-published benchmarks claim strong results on agentic, coding and browsing tasks.[1][2]
Limitations
- Headline benchmark figures come from Z.ai's own model card and OpenRouter's arena listing; independent verification is thin, so validate on your actual workload.[1][2]
- Text-only input and output; no vision or audio modality is reported in the supplied evidence.[1]
- Knowledge cutoff is unspecified, and self-hosting a 744B-parameter model demands substantial infrastructure: the vendor's serving examples use 8-way tensor parallelism.[1][2]
Where to use it
- Worth evaluating for long-horizon agentic coding and tool-use pipelines, an area both the vendor and arena data emphasize.[1][2]
- Candidate for long-document or large-codebase analysis given the reported ~200K-token context window; confirm effective recall on your own data.[1]
- Candidate for self-hosted deployments needing MIT-licensed weights and established serving frameworks; plan capacity for a very large MoE.[2]
Sources
- [1] OpenRouter catalog · Z.ai: GLM 5Sources checked: Nov 13, 2026
- [2] Model documentation · Z.ai: GLM 5Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
GLM-5 is Z.ai’s flagship open-source foundation model engineered for complex systems design and long-horizon agent workflows. Built for expert developers, it delivers production-grade performance on large-scale programming tasks, rivaling leading...
Specifications
- Model ID
- z-ai/glm-5
- Canonical ID
- z-ai/glm-5-20260211
- Creator
- z-ai
- Added to OpenRouter
- 11.02.2026
- Context length
- 204,800
- Modality
- text->text
- Input modalities
- text
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $1 / out $3.2
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | reasoning: optional · parameters: include_reasoning, reasoning |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: optional · parameters: include_reasoning, reasoning |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Design Arena | agents · androidnativewin rate 55% · rank 30 | 1,135Elo | CC — below the B band | 13.11.2026 |
| Design Arena | agents · fullstackwin rate 51.7% · rank 29 | 1,101Elo | CC — below the B band | 13.11.2026 |
| Design Arena | agents · godotgamedevwin rate 46.7% · rank 20 | 1,144Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | agents · htmlslideswin rate 45.1% · rank 20 | 1,044Elo | CC — below the B band | 13.11.2026 |
| Design Arena | agents · mobileappswin rate 51.7% · rank 37 | 1,130Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · 3dwin rate 56.3% · rank 47 | 1,212Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · asciiartwin rate 47.9% · rank 49 | 1,142Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · codecategorieswin rate 55.5% · rank 46 | 1,237Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · datavizwin rate 53% · rank 46 | 1,226Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · gamedevwin rate 57.4% · rank 45 | 1,225Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · svgwin rate 54.4% · rank 45 | 1,160Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 53.6% · rank 55 | 1,221Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · websitewin rate 55% · rank 51 | 1,240Elo | BB — next 30% of the cohort | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| StreamLakefp8 | 198,000 | $0.600 | $1.92 | — | — | 100% |
| GMICloudfp8 | 202,752 | $0.600 | $1.92 | — | — | 100% |
| Baidufp8 | 202,752 | $0.700 | $2.24 | — | — | 100% |
| SiliconFlowfp8 | 204,800 | $0.950 | $2.55 | — | — | — |
| Amazon Bedrock | 202,752 | $1 | $3.2 | — | — | 100% |
| Venicefp8 | 198,000 | $1 | $3.2 | — | — | 91.67% |
| Novitafp8 | 202,800 | $1 | $3.2 | — | — | — |
| Z.AIfp8 | 202,752 | $1 | $3.2 | — | — | 100% |
| Venice | 200,000 | $1.2 | $4 | — | — | — |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.