z-ai/glm-5.3-flash

模型发布 OpenRouter 模型 原文语言:中文

OpenRouter 于 2026-08-26 收录该模型(收录时间不等于模型正式发布日期);上下文:1,048,576 tokens;输入价:$0.0399/百万 tokens;输出价:$0.0784/百万 tokens;定位(来源描述):GLM-5.3-Flash is a native multimodal model from Z.ai.

全文

Z.ai: GLM 5.3 Flash

z-ai/glm-5.3-flash

Model weights

Compare

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 reducing compute overhead.

Modalities

In / out price

$0.0399 / $0.0784per 1M

Context

1.0M

Released

Aug 26, 2026

Compare

Providers

Different companies host the same model. OpenRouter routes your request to one of them based on the routing mode you pick: Balanced (price + speed), Nitro (fastest), or Floor (cheapest).

Latency / throughput

$0.0399$0.0784$0.03000.78s91 tok/s

100.00%

$0.0600$0.3000$0.03000.70s134 tok/s

99.91%

$0.1500$0.5000$0.03000.74s147 tok/s

100.00%

$0.0420$0.1840$0.04100.45s106 tok/s

100.00%

$0.1500$0.5000$0.03000.58s139 tok/s

100.00%

$0.1125$0.3750$0.03000.78s129 tok/s

100.00%

$0.1200$0.5000$0.04000.66s146 tok/s

100.00%

$0.1200$0.3000$0.03000.74s97 tok/s

99.77%

$0.1500$0.5000$0.03000.36s119 tok/s

100.00%

$0.1080$0.3500$0.03500.32s119 tok/s

100.00%

$0.1400$0.2400$0.03000.70s86 tok/s

100.00%

$0.1500$0.5000$0.03000.50s104 tok/s

100.00%

$0.1080$0.3500$0.03500.36s111 tok/s

100.00%

$0.1500$0.5000$0.03001.10s131 tok/s

99.77%

$0.0960$0.3200$0.01920.81s78 tok/s

99.97%

$0.0650$0.2400$0.06400.52s91 tok/s

100.00%

$0.1500$0.5000$0.03000.92s96 tok/s

99.97%

$0.1500$0.5000$0.03000.96s84 tok/s

100.00%

$0.1500$0.5000$0.03000.79s96 tok/s

100.00%

$0.1750$0.4700$0.04000.52s91 tok/s

99.97%

$0.1200$0.4500$0.03000.68s67 tok/s

100.00%

$0.1500$0.5000$0.03001.05s122 tok/s

99.22%

$0.0750$0.2500$0.01502.33s53 tok/s

99.68%

$0.1500$0.5000$0.03001.02s90 tok/s

99.83%

$0.0800$0.5000$0.01200.76s39 tok/s

99.97%

41% off

$0.15$0.0885$0.50$0.2950$0.03$0.01771.85s32 tok/s

99.86%

$0.1750$0.6250$0.02251.54s39 tok/s

100.00%

44% off

$0.15$0.0840$0.50$0.2800$0.03$0.01682.87s35 tok/s

100.00%

$0.0480$0.1380$0.02401.30s27 tok/s

100.00%

Fast

$0.1875$0.6250$0.03751.05s83 tok/s

99.67%

$0.1000$0.5000$0.03001.48s41 tok/s

99.77%

$0.1200$0.3800$0.02000.66s153 tok/s

96.30%

$0.1500$0.5000$0.03002.18s48 tok/s

100.00%

$0.1500$0.5000$0.03002.15s55 tok/s

100.00%

$0.1500$0.5000$0.03002.95s42 tok/s

98.27%

$0.1000$0.4000$0.03001.64s44 tok/s

99.15%

50% off

$0.15$0.0750$0.50$0.2500$0.03$0.01501.91s24 tok/s

99.80%

$0.2250$0.7500$0.04500.54s122 tok/s

94.49%

$0.1500$0.5000$0.03000.70s81 tok/s

99.44%

$0.0650$0.1800$0.01203.94s21 tok/s

96.98%

$0.1500$0.5000$0.03000.81s40 tok/s

99.96%

$0.1750$0.6250$0.02251.80s37 tok/s

100.00%

ProviderInput / Output /M

Input $0.0399 / Output $0.0784

91 tok/s100.00% uptime

Input $0.06 / Output $0.30

134 tok/s99.91% uptime

Input $0.15 / Output $0.50

147 tok/s100.00% uptime

Input $0.042 / Output $0.184

106 tok/s100.00% uptime

Input $0.15 / Output $0.50

139 tok/s100.00% uptime

Input $0.1125 / Output $0.375

129 tok/s100.00% uptime

Input $0.12 / Output $0.50

146 tok/s100.00% uptime

Input $0.12 / Output $0.30

97 tok/s99.77% uptime

Input $0.15 / Output $0.50

119 tok/s100.00% uptime

Input $0.108 / Output $0.35

119 tok/s100.00% uptime

Input $0.14 / Output $0.24

86 tok/s100.00% uptime

Input $0.15 / Output $0.50

104 tok/s100.00% uptime

Input $0.108 / Output $0.35

111 tok/s100.00% uptime

Input $0.15 / Output $0.50

131 tok/s99.77% uptime

Input $0.096 / Output $0.32

78 tok/s99.97% uptime

Input $0.065 / Output $0.24

91 tok/s100.00% uptime

Input $0.15 / Output $0.50

96 tok/s99.97% uptime

Input $0.15 / Output $0.50

84 tok/s100.00% uptime

Input $0.15 / Output $0.50

96 tok/s100.00% uptime

Input $0.175 / Output $0.47

91 tok/s99.97% uptime

Input $0.12 / Output $0.45

67 tok/s100.00% uptime

Input $0.15 / Output $0.50

122 tok/s99.22% uptime

Input $0.075 / Output $0.25

53 tok/s99.68% uptime

Input $0.15 / Output $0.50

90 tok/s99.83% uptime

Input $0.08 / Output $0.50

39 tok/s99.97% uptime

Input $0.0885 / Output $0.295

41% off

32 tok/s99.86% uptime

Input $0.175 / Output $0.625

39 tok/s100.00% uptime

Input $0.084 / Output $0.28

44% off

35 tok/s100.00% uptime

Input $0.048 / Output $0.138

27 tok/s100.00% uptime

Input $0.1875 / Output $0.625

Fast

83 tok/s99.67% uptime

Input $0.10 / Output $0.50

41 tok/s99.77% uptime

Input $0.12 / Output $0.38

153 tok/s96.30% uptime

Input $0.15 / Output $0.50

48 tok/s100.00% uptime

Input $0.15 / Output $0.50

55 tok/s100.00% uptime

Input $0.15 / Output $0.50

42 tok/s98.27% uptime

Input $0.10 / Output $0.40

44 tok/s99.15% uptime

Input $0.075 / Output $0.25

50% off

24 tok/s99.80% uptime

Input $0.225 / Output $0.75

122 tok/s94.49% uptime

Input $0.15 / Output $0.50

81 tok/s99.44% uptime

Input $0.065 / Output $0.18

21 tok/s96.98% uptime

Input $0.15 / Output $0.50

40 tok/s99.96% uptime

Input $0.175 / Output $0.625

37 tok/s100.00% uptime

Pricing

The average price customers actually pay for this model, next to the prices providers post. Caching and discounts mean the price actually paid is often well below the listed one.

Weighted average

Input

$0.04465/M tokens

Output

$0.39/M tokens

GLM 5.3 Flash — Price history

Price history

No effective price data for this selection

$0.0268$0.320590.3%3.7%

$0.0273$0.499577.4%2.0%

$0.0277$0.249878.8%0.5%

$0.0310$0.077889.8%7.3%

$0.0347$0.299784.2%5.6%

$0.0356$0.379684.4%4.3%

$0.0366$0.499690.6%1.6%

$0.0380$0.178650.9%0.0%

$0.0402$0.349192.8%6.0%

$0.0412$0.183373.2%3.6%

$0.0418$0.499690.2%9.0%

$0.0419$0.349390.6%2.0%

$0.0421$0.448986.6%0.1%

$0.0436$0.499688.6%2.4%

$0.0437$0.304266.5%3.1%

$0.0444$0.399179.3%0.3%

$0.0448$0.239086.5%2.8%

$0.0453$0.374381.4%4.1%

$0.0456$0.499687.0%10.1%

$0.0472$0.499785.7%1.7%

$0.0478$0.499690.2%7.9%

$0.0480$0.294257.2%0.2%

$0.0496$0.299178.2%0.7%

$0.0504$0.499683.0%6.5%

$0.0508$0.499682.7%1.2%

$0.0517$0.499881.9%0.1%

$0.0529$0.499780.9%1.4%

$0.0530$0.499680.9%0.5%

$0.0549$0.166575.1%0.6%

$0.0600$0.249824.9%0.1%

$0.0606$0.469384.7%1.3%

$0.0613$0.499073.9%0.1%

$0.0623$0.499673.1%0.1%

$0.0642$0.239270.5%0.4%

$0.0675$0.499768.7%1.7%

$0.0717$0.499865.3%5.0%

$0.0722$0.499364.8%0.0%

$0.0758$0.623965.0%0.0%

$0.0775$0.749281.9%1.6%

$0.0883$0.499651.4%0.3%

Fast

$0.0928$0.624763.1%0.2%

$0.1139$0.624440.0%0.0%

ProviderInput / Output /M

Token share

Effective in $0.0268 / Effective out $0.3205

Token share

3.7%

Effective in $0.0273 / Effective out $0.4995

Token share

2.0%

Effective in $0.0277 / Effective out $0.2498

Token share

0.5%

Effective in $0.0310 / Effective out $0.0778

Token share

7.3%

Effective in $0.0347 / Effective out $0.2997

Token share

5.6%

Effective in $0.0356 / Effective out $0.3796

Token share

4.3%

Effective in $0.0366 / Effective out $0.4996

Token share

1.6%

Effective in $0.0380 / Effective out $0.1786

Token share

0.0%

Effective in $0.0402 / Effective out $0.3491

Token share

6.0%

Effective in $0.0412 / Effective out $0.1833

Token share

3.6%

Effective in $0.0418 / Effective out $0.4996

Token share

9.0%

Effective in $0.0419 / Effective out $0.3493

Token share

2.0%

Effective in $0.0421 / Effective out $0.4489

Token share

0.1%

Effective in $0.0436 / Effective out $0.4996

Token share

2.4%

Effective in $0.0437 / Effective out $0.3042

Token share

3.1%

Effective in $0.0444 / Effective out $0.3991

Token share

0.3%

Effective in $0.0448 / Effective out $0.2390

Token share

2.8%

Effective in $0.0453 / Effective out $0.3743

Token share

4.1%

Effective in $0.0456 / Effective out $0.4996

Token share

10.1%

Effective in $0.0472 / Effective out $0.4997

Token share

1.7%

Effective in $0.0478 / Effective out $0.4996

Token share

7.9%

Effective in $0.0480 / Effective out $0.2942

Token share

0.2%

Effective in $0.0496 / Effective out $0.2991

Token share

0.7%

Effective in $0.0504 / Effective out $0.4996

Token share

6.5%

Effective in $0.0508 / Effective out $0.4996

Token share

1.2%

Effective in $0.0517 / Effective out $0.4998

Token share

0.1%

Effective in $0.0529 / Effective out $0.4997

Token share

1.4%

Effective in $0.0530 / Effective out $0.4996

Token share

0.5%

Effective in $0.0549 / Effective out $0.1665

Token share

0.6%

Effective in $0.0600 / Effective out $0.2498

Token share

0.1%

Effective in $0.0606 / Effective out $0.4693

Token share

1.3%

Effective in $0.0613 / Effective out $0.4990

Token share

0.1%

Effective in $0.0623 / Effective out $0.4996

Token share

0.1%

Effective in $0.0642 / Effective out $0.2392

Token share

0.4%

Effective in $0.0675 / Effective out $0.4997

Token share

1.7%

Effective in $0.0717 / Effective out $0.4998

Token share

5.0%

Effective in $0.0722 / Effective out $0.4993

Token share

0.0%

Effective in $0.0758 / Effective out $0.6239

Token share

0.0%

Effective in $0.0775 / Effective out $0.7492

Token share

1.6%

Effective in $0.0883 / Effective out $0.4996

Token share

0.3%

Effective in $0.0928 / Effective out $0.6247

FastToken share

0.2%

Effective in $0.1139 / Effective out $0.6244

Token share

0.0%

Performance

Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).

Throughput

153tok/s

P50, best across providers

Latency

0.32s

P50, best provider

Reasoning effortLatency / throughput

AutoExacto benchmarks

GPQA DiamondTAU-BenchDarkbloom92.5%--inference.net88.1%78.0%Z.ai88.2%77.4%DigitalOcean87.8%76.6%Baseten82.2%--inference.net (US)87.2%75.2%Baseten82.7%75.5%

+36 more providers

Uptime

Uptime is the percentage of the past 3 days that at least one provider was responding to requests. Availability is the percentage of time that inference was successfully served. OpenRouter continuously monitors and uses the next-best provider when one returns an error.

Uptime (3d)The model was reachable. Request routed to a provider.

100.00%

Availability (3d)The model returned inference from any provider. Errors and empty responses count against it.

99.77%

Availability over the last 3 days

Last 72 hours

Availability 99.77%

3 Days Ago2 Days AgoYesterdayNow

Availability over the last 24 hours

OpenRouter Availability

99.71%

Without Routing

97.17%

When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it. You can access per-provider uptime data programmatically through the Endpoints API. Learn more about our load balancing and provider routing options.

Benchmarks

Scores on standardized evaluations and this model’s own outputs on shared benchmark prompts. Higher percentages are better — and each rank shows where this model lands among the models evaluated on that source.

Benchmark score summary for Z.ai: GLM 5.3 Flash (Artificial Analysis, Design Arena and Vals AI)

SourceBenchmarkScore

Artificial AnalysisGLM-5.3-Flash Intelligence Index27.8

Artificial AnalysisGLM-5.3-Flash Coding Index71.5

Artificial AnalysisGLM-5.3-Flash Agentic Index50.9

Artificial AnalysisGLM-5.3-Flash GPQA Diamond91.2%

Artificial AnalysisGLM-5.3-Flash HLE39.9%

Artificial AnalysisGLM-5.3-Flash AA-LCR80.0%

Artificial AnalysisGLM-5.3-Flash τ-Bench Banking47.2%

Artificial AnalysisGLM-5.3-Flash GDPval-AA57.4%

Artificial AnalysisGLM-5.3-Flash CritPt15.4%

Artificial AnalysisGLM-5.3-Flash SciCode51.6%

Artificial AnalysisGLM-5.3-Flash Terminal-Bench 4.032.8%

Artificial AnalysisGLM-5.3-Flash Terminal-Bench 2.184.3%

Artificial AnalysisGLM-5.3-Flash AA-Omniscience Accuracy28.2%

Artificial AnalysisGLM-5.3-Flash AA-Omniscience Non-Hallucination Rate61.7%

Design ArenaGLM-5.3-Flash Models Arena 3D Elo1293

Design ArenaGLM-5.3-Flash Models Arena Asciiart Elo1256

Design ArenaGLM-5.3-Flash Models Arena Code Categories Elo1267

Design ArenaGLM-5.3-Flash Models Arena Data Visualization Elo1257

Design ArenaGLM-5.3-Flash Models Arena Game Development Elo1266

Design ArenaGLM-5.3-Flash Models Arena SVG Elo1270

Design ArenaGLM-5.3-Flash Models Arena UI Component Elo1294

Design ArenaGLM-5.3-Flash Models Arena Website Elo1257

Vals AIGLM 5.3 Flash Access to Justice Benchmark71.1%

Vals AIGLM 5.3 Flash Code Migration v1.151.6%

Vals AIGLM 5.3 Flash CyberBench v1.167.9%

Vals AIGLM 5.3 Flash CyberOffense18.3%

Vals AIGLM 5.3 Flash EMB v1.166.3%

Vals AIGLM 5.3 Flash Finance Agent v2.160.6%

Vals AIGLM 5.3 Flash Vals Finance Index55.8%

Vals AIGLM 5.3 Flash GPQA Diamond86.4%

Vals AIGLM 5.3 Flash HLAB-Vals-1.110.4%

Vals AIGLM 5.3 Flash IOI52.5%

Vals AIGLM 5.3 Flash LiveCodeBench80.5%

Vals AIGLM 5.3 Flash LegalBench83.9%

Vals AIGLM 5.3 Flash Vals Legal Index71.1%

Vals AIGLM 5.3 Flash Legal Research Bench45.2%

Vals AIGLM 5.3 Flash MedScribe88.9%

Vals AIGLM 5.3 Flash MMLU Pro86.1%

Vals AIGLM 5.3 Flash MMMU Pro86.0%

Vals AIGLM 5.3 Flash ProgramBench0.0%

Vals AIGLM 5.3 Flash ProofBench v1.121.0%

Vals AIGLM 5.3 Flash SAGE89.3%

Vals AIGLM 5.3 Flash SkillsBench40.2%

Vals AIGLM 5.3 Flash SRE Bench0.8%

Vals AIGLM 5.3 Flash SWE-bench92.0%

Vals AIGLM 5.3 Flash Tax Agent Bench31.6%

Vals AIGLM 5.3 Flash TaxEval v275.6%

Vals AIGLM 5.3 Flash Terminal-Bench 4.037.4%

Vals AIGLM 5.3 Flash Terminal-Bench-Science2.9%

Vals AIGLM 5.3 Flash Vals Index50.9%

Vals AIGLM 5.3 Flash Vibe Code Bench 1-10016.0%

Vals AIGLM 5.3 Flash Vibe Code Bench v1.130.8%

Apps

Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.

1.

Claude Code

Claude Code is Anthropic's agentic coding tool that reads your entire codebase, plans and executes changes across files, runs tests, and iterates on failures, all from natural language prompts.

6.83Ttokens

2.

Hermes Agent

Hermes Agent is an open-source, self-improving AI agent by Nous Research that runs persistently with memory across sessions, and builds reusable skills from experience. It comes with 40+ built-in tools, including web search, browser automation, and vision, plus scheduled automations and subagents.

3.45Ttokens

3.

Portkey AI

Control panel for AI apps

3.37Ttokens

4.

pi

There are many coding agents, but this one is yours.

1.48Ttokens

5.

Lemonade

The AI tool for Roblox games.

1.13Ttokens

Rest of OpenRouter

377B

Top 5 Apps

189B

Activity

Token volume and request traffic to this model over time.

API

Drop-in code to call this model. OpenRouter's API is OpenAI-compatible — most SDKs work by just swapping the base URL. The only thing that changes between models is the model slug below.

Z.ai: GLM Flash versions

Available through EU in-region routing

Z.ai: GLM 5.3 Flash is available through EU in-region routing. Requests sent to eu.openrouter.ai are decrypted in the EU and routed only to shared provider endpoints in Europe (the EU/EEA, the UK, and Switzerland).

Providers in EuropeInceptronPrices in Europe

Inceptron: $0.175/M tokens input, $0.47/M tokens output

Base URLhttps://eu.openrouter.ai/api/v1PlanRequires Business or Enterprise. Self-serve on Business: no contract, no monthly minimum. OpenRouter Business for EU in-region routing

Fails closed: if no shared provider endpoint in Europe is available, the request errors instead of leaving the region. Requests served by your own BYOK key go to your deployment’s region, which you control. In-region routing docs

Documentation

Guides, cookbooks, blog posts, and tools for building with Z.ai: GLM 5.3 Flash on OpenRouter.

Open-Weight vs Closed-Weight Models: What's the Difference?Blog post

Frequently asked questions

Yes, the stealth model Ox Alpha was revealed to be ZAI's new model, GLM-5.3 Flash.

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 reducing compute overhead.

The GLM 5.3 Flash documentation on OpenRouter includes the Open-Weight vs Closed-Weight Models: What's the Difference? blog post. The quickstart covers API keys and your first request, and the API reference lists every parameter.

GLM 5.3 Flash costs $0.0399/M input tokens and $0.0784/M output tokens, with separate rates for Cache Read at $0.03/M tokens.

GLM 5.3 Flash has a 1,048,576 token context window. It supports up to 131,072 completion tokens.

Yes. GLM 5.3 Flash accepts tools for function calling on 36 of the 37 providers serving it, and requests that send tools are routed to those providers. It also supports structured outputs via a JSON schema in response_format.

GLM 5.3 Flash accepts text, images, and video as input and returns text.

GLM 5.3 Flash is served by 37 providers on OpenRouter: Relace, Wafer, Morph, Open Inference, Darkbloom, Tenstorrent, GMICloud, NEAR AI and 29 more. Requests are routed to the best available provider, with automatic failover to the others, and you can pin or exclude providers with provider routing.

GLM 5.3 Flash was released on August 26, 2026.

正文以来源语言显示。

参考来源

← 返回每日简报