z-ai
Z.ai: GLM 5.3 Flash
z-ai/glm-5.3-flash
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Z.ai’s GLM-5.3-Flash is described in its documentation as a natively multimodal GLM-5 model with 320B total and 18B active parameters, accepting text, image, and video inputs and producing text. The OpenRouter catalog lists a 1,048,576-token context window, tool and structured-output support, and mandatory reasoning with low, high, and max effort settings. OpenRouter also reports benchmark measurements across design, coding, agentic, and intelligence categories, while Z.ai positions the model for efficient coding and long-horizon agent tasks; validate these claims on real workloads.[1][2]
Strengths
- Accepts text, image, and video inputs and emits text; Z.ai calls it the first natively multimodal model in the GLM-5 series.[1][2]
- Offers a 1,048,576-token context window; Z.ai says its hybrid sparse and linear attention is designed to retain precise long-context behavior while reducing serving cost.[1][2]
- The documentation declares an MIT license and lists SGLang, vLLM, Transformers, KTransformers, Unsloth, and TokenSpeed deployment paths; the catalog also lists tool and structured-output support.[1][2]
Limitations
- The evidence pack contains an OpenRouter catalog entry and Z.ai’s own documentation, so vendor performance and efficiency claims still need independent, workload-specific validation.[1][2]
- OpenRouter marks reasoning as mandatory and max as the default; Z.ai says low or high must be requested explicitly, so applications should confirm the required reasoning behavior.[1][2]
- The sources do not provide a knowledge cutoff, hardware requirements, or safety and guardrail details; benchmark footnotes also use specific sampling and context settings that may not match production use.[1][2]
Where to use it
- Consider for coding assistants, repository-scale analysis, and agent workflows that need tools; pilot on your own tasks before rollout.[1][2]
- Consider for multimodal applications that must inspect text, images, or video and return text.[1][2]
- Consider for long-context document or media analysis where a very large context window matters; measure accuracy and cost in the target setting.[1][2]
Sources
- [1] OpenRouter catalog · Z.ai: GLM 5.3 FlashSources checked: Nov 13, 2026
- [2] Model documentation · Z.ai: GLM 5.3 FlashSources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
GLM-5.3-Flash is a native multimodal model from Z.ai. It is suited for efficient coding and long-horizon agent tasks. Its hybrid sparse and linear attention architecture maintains accurate long-context behavior while...
Specifications
- Model ID
- z-ai/glm-5.3-flash
- Canonical ID
- z-ai/glm-5.3-flash-20260826
- Creator
- z-ai
- Added to OpenRouter
- 26.08.2026
- Context length
- 1,048,576
- Modality
- text+image+video->text
- Input modalities
- text, image, video
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.040 / out $0.078
- 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 | BB — next 30% of the cohort | reasoning: mandatory · default max · efforts: max, high, low · parameters: include_reasoning, reasoning, reasoning_effort |
| Coding | AA — next 20% of the cohort | — |
| Agents | SS — top 10% of the cohort | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: mandatory · default max · efforts: max, high, low · parameters: include_reasoning, reasoning, reasoning_effort |
| Document analysis | No measurement | — |
| Vision | No measurement | input: image |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | input: video |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Artificial Analysis | Intelligence index | 27.8 | BB — next 30% of the cohort | 13.11.2026 |
| Artificial Analysis | Coding index | 71.5 | AA — next 20% of the cohort | 13.11.2026 |
| Artificial Analysis | Agentic index | 50.9 | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | models · 3dwin rate 55% · rank 20 | 1,296Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | models · asciiartwin rate 54% · rank 11 | 1,269Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | models · codecategorieswin rate 48.3% · rank 28 | 1,268Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | models · datavizwin rate 50.2% · rank 33 | 1,257Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | models · gamedevwin rate 46.5% · rank 27 | 1,268Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | models · svgwin rate 54.8% · rank 15 | 1,272Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 52.1% · rank 19 | 1,298Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | models · websitewin rate 47.3% · rank 39 | 1,258Elo | AA — next 20% 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) |
|---|---|---|---|---|---|---|
| Relacefp4 | 1,048,576 | $0.040 | $0.078 | — | — | 100% |
| Wafernvfp4 | 1,048,576 | $0.042 | $0.184 | — | — | 100% |
| OpenInferencefp4 | 1,048,576 | $0.060 | $0.300 | — | — | 100% |
| Darkbloomfp4 | 262,144 | $0.065 | $0.180 | — | — | 90% |
| Wafernvfp4 | 1,048,576 | $0.065 | $0.240 | — | — | 100% |
| Tenstorrentfp8 | 262,000 | $0.075 | $0.250 | — | — | 100% |
| GMICloudfp8 | 1,048,576 | $0.075 | $0.250 | — | — | 100% |
| Near AIfp8 | 1,048,576 | $0.080 | $0.500 | — | — | 100% |
| Novitafp8 | 1,048,576 | $0.084 | $0.280 | — | — | 100% |
| StreamLakefp8 | 1,024,000 | $0.089 | $0.295 | — | — | 100% |
| SiliconFlowfp8 | 1,048,576 | $0.100 | $0.400 | — | — | 100% |
| DekaLLM | 1,048,576 | $0.100 | $0.500 | — | — | 100% |
| InferenceNetfp4 | 1,048,576 | $0.108 | $0.350 | — | — | 100% |
| InferenceNetfp4 | 1,048,576 | $0.108 | $0.350 | — | — | 100% |
| DigitalOcean | 1,048,576 | $0.112 | $0.375 | — | — | 100% |
| Morphfp8 | 1,048,576 | $0.116 | $0.122 | — | — | 100% |
| Reka | 1,048,576 | $0.120 | $0.300 | — | — | 100% |
| CoreWeavenvfp4 | 1,048,576 | $0.120 | $0.380 | — | — | 100% |
| River | 1,048,576 | $0.120 | $0.400 | — | — | 100% |
| NextBitfp4 | 1,048,576 | $0.120 | $0.450 | — | — | 100% |
| Particle AInvfp4 | 1,048,576 | $0.120 | $0.500 | — | — | 100% |
| Makorafp4 | 1,048,576 | $0.140 | $0.240 | — | — | 100% |
| Mistral | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| KT | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Coherenvfp4 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Nebiusfp4 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Modalfp4 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Crusoefp4 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| AtlasCloudfp8 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Fireworks | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Friendli | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Together | 1,048,575 | $0.150 | $0.500 | — | — | 100% |
| Parasailfp4 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| BaseTenfp8 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Venice | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| DeepInfrafp4 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Z.AIfp8 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Inceptronfp8 | 1,048,576 | $0.175 | $0.470 | — | — | 100% |
| Sail Researchfp4 | 1,048,576 | $0.175 | $0.625 | — | — | 100% |
| Sail Researchfp4 | 1,048,576 | $0.175 | $0.625 | — | — | — |
| Parasailfp4 | 1,048,576 | $0.188 | $0.625 | — | — | 100% |
| Fireworks | 1,048,576 | $0.225 | $0.750 | — | — | 96.67% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.