Scan to join WeChat groupGLM-5.3-FlashX
GLM-5.3-FlashX for faster multimodal and agent workloads
GLM-5.3-FlashX accepts text, image, video and file input with a 1M-token context window, 128K maximum output and an official rated speed of up to 200 tokens/s.
starsCapabilities
paymentsContext and pricing
descriptionOverview
Overview
GLM-5.3-FlashX is the faster GLM-5.3-Flash variant, using model ID glm-5.3-flashx. Official documentation lists a 1,048,576-token context window, 131,072-token maximum output, and text, image, video and file input. Zhipu rates generation speed at up to 200 tokens/s.
Capabilities and parameters
The model supports native visual understanding, streaming, function calling, context caching and structured output. thinking.type only supports enabled. Zhipu recommends temperature: 1, top_p: 0.95 and reasoning_effort: max; streaming tool calls should also enable tool_stream.
API pricing
Zhipu lists API pricing at CNY 2 input, CNY 7 output and CNY 0.57 cache read per million tokens. Cache storage is temporarily free. Actual billing follows the pricing in effect when the request runs.
lightbulbUse cases
- Low-latency multimodal chat
- Visual coding and UI recreation
- Long-context agents and tool use
- Document, finance and Office workflows
thumb_upStrengths
- Official rated speed up to 200 tokens/s
- 1M context and 128K maximum output
- Native text, image, video and file input
- Tool calling, caching and structured output
infoLimitations
- Reasoning cannot be disabled
- Actual speed depends on request size and service load
- FlashX is not currently available in Coding Plan
- Pricing and capabilities may change