CrossModel
GLM-5.3
z-ai/glm-5.3
Back to model catalog

Z.ai · Model guide

GLM-5.3

Compare
z-ai/glm-5.3

Z.AI's latest flagship for complex software engineering, long-running agents and cybersecurity work. It keeps GLM-5.2's 1M-token base model while using expanded post-training to improve end-to-end project delivery, terminal work and vulnerability analysis.

Modalities
TextText
Context
1M
Max output
128K
Price / 1M tokens
$ 1.20in$ 4.40out
StreamingTool callingJSON outputImage inputThinking mode

What this model is good at

What the vendor positions it for, and which of our models to reach for instead.

VendorA flagship foundation model with major advances in complex software engineering and agent tasks.source

Best for

  • Long-horizon software engineering · Official evaluations emphasize terminal work, repository-scale changes and complete delivery across complex engineering environments.(vendor claim)source

  • Security-focused code review · The release highlights white-box vulnerability discovery and validation across CyberGym, ExploitBench and ExploitGym.source

Neighbouring models
  • z-ai/glm-5.2predecessor —Use GLM-5.2 when compatibility with applications that can disable thinking is required.
Compare them side by side →

Pricing and billing

Billed per token. Cached input is charged at the cache-read rate.

Price / 1M tokensYou pay
Input$ 1.20
Output$ 4.40
Cache read$ 0.30
Cache write$ 1.20
Estimate a request
$ 0.0208
Estimated cost per request
10.0K × $ 1.20
+ 2.0K × $ 4.40
≈ $ 20.80 per 1,000 requests

Capabilities and limits

What CrossModel guarantees across every route this model can take right now.

Context window
1.0M tokens
Max output
128.0K tokens
Input / output modalities
Text → Text
Streaming
Supported
Tool calling
Supported
Structured output (JSON)
Supported
Image input
Not supported
Thinking mode
Supported· always on
Reasoning effort
low · high · max
Available endpoints
/v1/chat/completions · /v1/responses · /v1/messages
First-party sources:Specsdocs.bigmodel.cn 1z.ai 2Capabilitiesdocs.bigmodel.cn

Published benchmarks

Scores the vendor reported, with the evaluation setup each one came from.

Agentic coding
  • Terminal-Bench 3.0
    pass_rate·Z.AI · 2026-08-14effort maxClaude Code 2.1.2073 runssetup not fully disclosed

    avg@3 over three rollouts per task; 400K context, Tool Search disabled, 600-turn cap and 10-hour timeout.

    28.3%
Agentic work
  • Agents' Last Exam (CLI)
    score·Z.AI · 2026-08-14effort maxsetup not fully disclosed
    28.5%
Coding
  • DeepSWE 1.1
    resolved_rate·Z.AI · 2026-08-14mini-swe-agentsetup not fully disclosed

    Official run used a six-hour timeout and 400K context.

    66.9%
Cybersecurity
  • CyberGym
    score·Z.AI · 2026-08-14effort maxClaude Code 2.1.207setup not fully disclosed

    Single-run Pass@1 over 1,507 tasks with no web tools and unlimited per-task timeout.

    84.5%
  • ExploitBench
    score·Z.AI · 2026-08-14effort maxClaude Code 2.1.2073 runssetup not fully disclosed

    Average coverage over 41 tasks across three revisions, with no web tools and a 300-interaction cap.

    54.4%
  • ExploitGym (6h)
    tasks_completed·Z.AI · 2026-08-14effort maxClaude Code 2.1.207setup not fully disclosed

    Single-run Pass@1 over 869 tasks under a six-hour time-normalized budget; no web tools; results rescaled at 115 TPS plus non-API overhead.

    130
Knowledge work
  • GDPval-AA v2
    elo·Z.AI · 2026-08-14setup not fully disclosed

    Artificial Analysis evaluation reproduced in Z.AI's official release comparison.

    1769
Reasoning
  • Humanity's Last Exam (with tools)
    accuracy·Z.AI · 2026-08-14effort maxsetup not fully disclosed

    The official comparison labels this HLE w/ Tools but does not disclose the complete tool set.

    62.5%

Vendor-reported numbers, not CrossModel measurements. Scores are only comparable when the benchmark version, metric and evaluation setup match, so nothing here is averaged or ranked.

Use it in your tools

Point the base URL at CrossModel and paste this model ID — every tool below has a setup guide.

Claude Code
Anthropic's official command-line coding assistant
Cursor
AI-first code editor
Cline
Open-source AI coding agent for VS Code
Codex
OpenAI's coding agent for ChatGPT desktop and the terminal
OpenCode
Open-source terminal coding assistant
Zed
High-performance, AI-native code editor
Open WebUI
Open-source, self-hosted AI chat interface
Continue
Open-source AI coding assistant for VS Code & JetBrains
See all integration guides →

Frequently asked questions

What is GLM-5.3?
GLM-5.3 is available on CrossModel as z-ai/glm-5.3. It has a 1M-token context window and can return up to 128K tokens per request.
How much does GLM-5.3 cost?
Currently $ 1.20 per 1M input tokens and $ 4.40 per 1M output tokens. Cached input is billed at $ 0.30 per 1M.
Does GLM-5.3 support tool calling and structured output?
Tool calling: Supported · Structured output (JSON): Supported · Image input: Not supported · Streaming: Supported · Thinking mode: Supported
Which endpoint do I call?
For text models, OpenAI Chat Completions (/v1/chat/completions), OpenAI Responses (/v1/responses) and Anthropic Messages (/v1/messages) all work — pick whichever your SDK already speaks.
Can I try it without writing code?
Yes. Sign in and open the Playground on this page to chat with the model directly — no API key needed. Usage is billed from your wallet at the normal rate.

Chat first, integrate later

Chat first, then wire it in once you like the answers.