GLM 5.2
GLM 5.2 is a large-scale reasoning model from Z.ai.
glm-5.2Released 16 Jun 2026 (Jun 2026)Pricing
List prices at Z.ai, in US dollars per million tokens.
| Tier | Input / 1M | Output / 1M | |
|---|---|---|---|
| Standard | $1.40 | $4.40 | |
| Cached input 81% off input | $0.26 | 81% off input | |
| Cache write, 5 minutes | $0.00 |
What it costs a month
At Z.ai list prices, 30 days.
- Chat assistant5,000 requests a day, 2K in / 0.5K out, cached$647
- RAG over documents2,000 requests a day, 8K in / 0.6K out, cached$557
- Coding agent1,000 requests a day, 40K in / 2K out, cached$986
- Overnight batch20,000 requests a day, 4K in / 0.4K out$4,416
Where to run it
27 hosts serve GLM 5.2. The comparison is on blended price, 3 parts input to 1 part output.
| Host | In / out per 1M | Model id | |
|---|---|---|---|
Z.aiLab API glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
CLF AI GatewayCloud glm-5.2 | $0.56 / $1.76 −60% vs headline | glm-5.2 | |
SCX.aiCloud GLM-5.2 | $0.61 / $1.98 −56% vs headline | GLM-5.2 | |
W&B InferenceInference GLM-5.2 | $0.62 / $2.42 −50% vs headline | GLM-5.2 | |
DeepInfraInference GLM-5.2 | $0.75 / $2.40 −46% vs headline | GLM-5.2 | |
Kyma APICloud glm-5.2 | $0.903 / $3.08 −33% vs headline | glm-5.2 | |
ai&Cloud glm-5.2 | $1.00 / $4.00 −19% vs headline | glm-5.2 | |
DatabricksCloud databricks-glm-5-2 | $1.40 / $4.40 Retires 4 Feb 2027 | databricks-glm-5-2 | Retires 4 Feb 2027 |
Mistral AILab API glm-5-2 | $1.40 / $4.40 | glm-5-2 | |
Alibaba DashScopeLab API glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
Qianwen AI PlatformLab API glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
Qwen CloudLab API glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
EmpirioLabs AICloud glm-5-2 | $1.40 / $4.40 | glm-5-2 | |
glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
OpenCode GoCloud glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
StreamLakeCloud glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
Tencent TokenHubCloud glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
Vertex AICloud glm-5.2-maas | $1.40 / $4.40 | glm-5.2-maas | |
BasetenInference GLM-5.2 | $1.40 / $4.40 | GLM-5.2 | |
FriendliAIInference GLM-5.2 | $1.40 / $4.40 | GLM-5.2 | |
Nebius AI StudioInference GLM-5.2 | $1.40 / $4.40 | GLM-5.2 | |
Together AIInference GLM-5.2 | $1.40 / $4.40 | GLM-5.2 | |
GMI CloudInference GLM-5.2-FP8 | $1.40 / $4.40 | GLM-5.2-FP8 | |
Novita AIInference glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
Fireworks AIInference glm-5p2 | $1.40 / $4.40 | glm-5p2 | |
PerplexityGateway glm-5.2 | $1.40 / $4.40 | glm-5.2 | |
ScalewayCloud glm-5.2 | $1.80 / $5.50 +27% vs headline | glm-5.2 |
Show 1 more listings: regions, snapshots and variantsHide regional listings
| Fireworks AI | |||||||
| Alias glm-5p2 | $1.40 / $4.40 | glm-5p2 | |||||
Capabilities
Core
- Streaming
- Function calling
- JSON mode
- Structured outputs
Multimodal
- Vision
- Image input
- Audio
- Image generation
- Video
Advanced
- Reasoning
- Tool calling
- MCP
- Prompt caching
- Batch API
- Embeddings
- Takes in
- Text
- Produces
- Text
- Features
- Prompt caching
- Open weights
- License: MITRuns in about 1808 GB of GPU memory
Cheaper alternatives
Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.
Same model, cheaper host
- CLF AI Gateway60% lessglm-5.2$0.56 / $1.76
- SCX.ai56% lessGLM-5.2$0.61 / $1.98
- W&B Inference50% lessGLM-5.2$0.62 / $2.42
- DeepInfra46% lessGLM-5.2$0.75 / $2.40
- Kyma API33% lessglm-5.2$0.903 / $3.08
Smaller Zhipu AI models
- 12% less
Other frontier labs
- 7% less
- 30% less
- 42% less
- 91% less
Open-weight picks
- 7% less
- 20% less
- 60% less
- 63% less
Model line
Newer and older versions, at each lab's own price.
Call it
Using Z.ai's API.
from litellm import completion
response = completion(
model="zai/glm-5.2",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)Quick answers
How much does GLM 5.2 cost?
On Z.ai, GLM 5.2 costs $1.40 per million input tokens and $4.40 per million output tokens. Cached input costs $0.26 per million tokens. The cheapest host is CLF AI Gateway, at $0.56 per million input tokens and $1.76 per million output tokens.
Which APIs offer GLM 5.2?
GLM 5.2 runs on 27 APIs: Z.ai, CLF AI Gateway, SCX.ai, W&B Inference, DeepInfra, Kyma API, ai&, Databricks, Mistral AI and 18 more.
What is GLM 5.2's context window?
1M tokens, with up to 128K tokens in one reply.
Is GLM 5.2 being retired?
Databricks retires it on 4 Feb 2027. 26 hosts keep serving it.
Is there a newer version of GLM 5.2?
Yes. GLM 5.3 is the newest model in this line, released on 18 Aug 2026.
What are cheaper alternatives to GLM 5.2?
Models that cost less and keep most of what it can do: GLM 5v Turbo on Z.ai (12% less), GPT-6.1 Terra on OpenAI (7% less), Gemini 3.8 Flash on Google AI Studio (30% less) and Muse Spark Open on Baseten (7% less). GLM 5.2 itself costs 60% less on CLF AI Gateway. Savings are on a mix of 3 input tokens to 1 output token.
When was GLM 5.2 released?
Zhipu AI released it on 16 Jun 2026.
What is GLM 5.2's API model id?
Use glm-5.2 with Z.ai. Other hosts use their own ids: GLM-5.2 on SCX.ai, GLM-5.2 on W&B Inference, GLM-5.2 on DeepInfra, databricks-glm-5-2 on Databricks.
Sources
- Baseten pricing
- DeepInfra pricing
- FriendliAI pricing
- Nebius AI Studio pricing
- SCX.ai pricing
- Together AI pricing
- W&B Inference pricing
- GMI Cloud pricing
- Databricks pricing
- EmpirioLabs AI pricing
- Mistral AI pricing
- ai& pricing
- CLF AI Gateway pricing
- Alibaba DashScope pricing
- Novita AI pricing
- OpenCode Go pricing
- Perplexity pricing
- Qianwen AI Platform pricing
- Qwen Cloud pricing
- Scaleway pricing
- StreamLake pricing
- Tencent TokenHub pricing
- Z.ai pricing
- Vertex AI pricing
- Fireworks AI pricing
- LiteLLM model map
Raw catalog entry for glm-5.2 on Z.ai
{
"cache_creation_input_token_cost": 0,
"cache_read_input_token_cost": 2.6e-7,
"input_cost_per_token": 0.0000014,
"litellm_provider": "zai",
"max_input_tokens": 1000000,
"max_output_tokens": 128000,
"mode": "chat",
"output_cost_per_token": 0.0000044,
"source": "https://docs.z.ai/guides/overview/pricing",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": true
}