Zhipu AI: GLM-5.3
glm-5.3- Input / Output
- $1.40 / $4.40
- Context
- 1M
- Released
- Aug 14, 2026
- Max output
- 128K
- Modalities
- Chat
- Capabilities
- Tool usePrompt CacheReasoning
About GLM-5.3
GLM-5.3 is Z.ai's flagship model for agentic coding and complex software projects, using the GLM-5.2 base with new post-training. Reasoning is always on, with low, high, and max effort levels. It is a text-only model that supports structured output and function calling, and Z.ai reports about 50 percent better results than GLM-5.2 on its Code Bench.
Where it works well
- Z.ai reports a 50 percent gain over GLM-5.2 on its own Z.ai Code Bench.
- Three effort levels let a team trade cost for depth, with max recommended for complex coding.
- At max effort Z.ai reports 34.5 percent on a task set at about 75K output tokens, against 23.4 percent at 96K for GLM-5.2.
- Terminal-Bench 3.0 improves from 4.6 to 28.3 by Z.ai's figures, and CyberGym reaches 84.5 percent.
When to choose another model
- Reasoning cannot be turned off, so quick low-stakes calls cost more latency than on a non-reasoning model.
- Input is limited to text; there is no image or file reading.
Getting started
Create API key
Create a key in Console, then use it with every model on the platform.
Send your first request
Copy the example for your language and run it against the endpoint.
Text to textResponses APIPOST/v1/responsesAPI format:curl https://api.tokenlab.sh/v1/responses \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxx" \ -d '{ "model": "glm-5.3", "input": "Hello!" }'
Pricing
The Verified price applies to Verified, which costs less on most models. The Official price is the model maker's published price and applies to the more reliable Official route. Auto bills the route that completes the request.
Default
per 1M tokens- Official price
- Input $1.40 / Output $4.40 / Cache read $0.26
- Verified price
- —
- Discount
- —
Cache read
- Official price
- $0.26
- Verified price
- —
- Discount
- —
| Official priceper 1M tokens | Verified priceper 1M tokens | Discount | |
|---|---|---|---|
| Default | Input $1.40 / Output $4.40 / Cache read $0.26 | — | — |
| Prompt cache pricing | |||
| Cache read | $0.26 | — | — |
Usage & activity
Success rate is the share of requests that completed. Latency is how long a full response takes; P95 means 95% of requests finished within that time.
Usage & availability
Last 24 hours- Requests
- Success rate
- P95 latency
- Total tokens
- 30-day success rate
- 100.0%
- 30-day requests
- 100+
- Last active
- yesterday
Data is based on aggregate user requests, excluding status checks.
Open in Console
Open GLM-5.3 in Console with a prompt ready to edit or send.
Help me try glm-5.3 with a short message at /v1/responses. Show the reply, latency, and cost.
Use cases
Best for- Code
- Reasoning
Hard coding agents
Run it behind a coding agent for multi-file features, bug hunts, and refactors, using max effort for the difficult ones.
Security analysis
Review code for vulnerabilities and reproduce issues, an area where Z.ai reports its CyberGym result.
Terminal automation
Automate shell-heavy workflows such as environment setup, build repair, and deployment scripting.
Prompt examples
Reproduce this memory-safety bug from the report, write a failing test, then patch it and show the test passing.
Using low effort, rename this config key across the repo and update the docs.
Migrate our job queue from cron scripts to a worker service, and explain each step before you run it.
This model has conditional pricing. Monthly token totals alone cannot produce a reliable estimate; use the detailed pricing for the request specification, cache, and applicable time window.
FAQ
What is new in GLM-5.3?
It shares the GLM-5.2 base model, and every gain comes from post-training. Z.ai reports stronger agentic coding at each effort level with fewer output tokens, plus structured output and function calling.
What do the effort levels mean?
Low, high, and max set how much the model reasons. Low is lightweight, high is deeper, and max is deepest and is recommended for complex coding. Z.ai says reasoning can no longer be disabled.
What input does GLM-5.3 take?
Text only. GLM-5.3 reads prompts, source code, logs, and documents as plain text and replies with text, so the whole task needs to be expressible in text.
Should I choose GLM-5.3 or GLM-5.3-Flash?
GLM-5.3 is the flagship for hard coding and agent tasks. GLM-5.3-Flash is natively multimodal and far cheaper, with Z.ai reporting 63.4 on DeepSWE v1.1; test it first if images matter or volume is high.
How much does GLM-5.3 cost?
On TokenLab, GLM-5.3 costs Input $1.40 / Output $4.40 per 1M tokens. The pricing table above shows the full breakdown. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.
What are the context window and output limit of GLM-5.3?
GLM-5.3 accepts up to 1,000,000 tokens of context and returns up to 128,000 tokens in one response.
Which endpoint should GLM-5.3 use?
Use https://api.tokenlab.sh/v1/responses for GLM-5.3. The request example below shows the matching code shape.
Which operations does GLM-5.3 support?
GLM-5.3 supports Text to text. Select an operation above to see its endpoint and request example.
Compare GLM-5.3
Sources
Reviewed Oct 2, 2026
