GLM 5.3 Flash is Z.ai's long-context chat model behind an OpenAI-compatible call. It reads up to 1,048,576 tokens, replies up to 131,072, and offers three reasoning-effort levels, 128 tool schemas per request and JSON output.
Workbench
These examples were written for this endpoint. Replace the prompt and sample media URLs with your own accessible inputs, then check the current quote and model availability before submitting.
- Input types
- Text
- Output types
- Text
- Input
- $0.1275 / 1 million tokens
- Cached input
- $0.0255 / 1 million tokens
- Output
- $0.425 / 1 million tokens
- Context
- 1,048,576
- Max output
- 131,072
$0.128/0.425per 1 million tokens · input / output
source-review A long read with a short, structured answer - the shape this window is built for.
- 6000
- 0.2
- high
- Input types
- Text
- Output types
- Text
- Input
- $0.1275 / 1 million tokens
- Cached input
- $0.0255 / 1 million tokens
- Output
- $0.425 / 1 million tokens
- Context
- 1,048,576
- Max output
- 131,072
$0.128/0.425per 1 million tokens · input / output
million-token-digest The corpus case: thousands of tickets in one call, with the model asked for patterns rather than a summary of each.
- 12000
- 0
- high
- Input types
- Text
- Output types
- Text
- Input
- $0.1275 / 1 million tokens
- Cached input
- $0.0255 / 1 million tokens
- Output
- $0.425 / 1 million tokens
- Context
- 1,048,576
- Max output
- 131,072
$0.128/0.425per 1 million tokens · input / output
json-brief response_format plus temperature 0, which is what makes the reply safe to hand to a parser.
- 16000
- 0
- high
This example starts at Input $0.1275 / Output $0.425 per 1 million tokens. The Playground updates the estimate when you change the settings.