GLM-5.2
A 744B-parameter MoE with 40B active, MIT-licensed, and tuned for long-horizon coding work. It sits between DeepSeek and Kimi on price and is the usual challenger when a cheaper model keeps looping or missing repository conventions — enough capability to finish the task without flagship pricing.
Pricing checked against provider documentation on . How we verify
glm-5.2 reasoning open weights Input
$1.40
/ 1M tokens
Output
$4.40
/ 1M tokens
Context
1M
1,000,000 tokens
Max output
131K
tokens per request
Pricing caveat: Independently measured by Artificial Analysis; the vendor's own pricing page was unreachable at time of checking. Treat as approximate and confirm in the console.
What it costs in practice
A workload of 10M input and 2M output tokens per month — roughly a moderately busy production assistant — runs $22.80/month on GLM-5.2. The next cheaper option, Claude Haiku 4.5 , would cost $20.00.
Where GLM-5.2 fits
- Coding agents on large repositories
- Long-horizon refactors
- Self-hosted agent stacks
- Structured output pipelines
Specifications
- API model ID
- glm-5.2
- Provider
- Z.ai
- Model family
- GLM-5
- Released
- July 2026
- Knowledge cutoff
- February 2026
- Context window
- 1,000,000 tokens
- Max output
- 131,072 tokens
- Native reasoning
- Yes
- Relative latency
- medium
- Input modalities
- text, image
- Output modalities
- text
- Open weights
- MIT
- Cached input
- $0.26 / 1M tokens
- Batch discount
- —
Frequently asked
How much does GLM-5.2 cost?
GLM-5.2 costs $1.40 per million input tokens and $4.40 per million output tokens, with cached input at $0.26.
What is GLM-5.2's context window?
GLM-5.2 accepts up to 1,000,000 tokens of context and can generate up to 131,072 output tokens per request.
Is GLM-5.2 a reasoning model?
Yes. GLM-5.2 performs native chain-of-thought before answering. Those thinking tokens are billed at the output rate, so budget above the sticker price.
What is GLM-5.2 best for?
MIT-licensed coding agents. A 744B-parameter MoE with 40B active, MIT-licensed, and tuned for long-horizon coding work. It sits between DeepSeek and Kimi on price and is the usual challenger when a cheaper model keeps looping or missing repository conventions — enough capability to finish the task without flagship pricing.