GLM-5.3-FlashX
Overview
GLM-5.3-FlashX is Z.AI's speed-optimized variant of GLM-5.3-Flash. It combines native multimodal understanding with a 1M-token context window for interactive coding, tool-using agents, and document workflows. Z.AI reports generation speeds of up to 200 tokens per second; actual speed varies by request and is not guaranteed.
Model ID: glm-5.3-flashx
Key Features
- Accelerated Inference: Emphasizes faster generation through serving-infrastructure improvements, helping reduce wait time in interactive and tool-driven workflows.
- Efficient Architecture: Uses the shared Flash foundation with 320 billion total parameters and 18 billion activated parameters, combining sparse and linear attention to reduce long-context computation.
- Native Visual Understanding: Accepts images, videos, text, and files for tasks that combine visual evidence with language and code.
- Configurable Reasoning: Supports
low,high, andmaxreasoning effort, withmaxas the default. Thinking remains enabled. - Agent Integration: Supports function calling, streamed tool calls, JSON output, and automatic context caching.
Best Use Cases
- Interactive Coding Agents: Implementation and debugging loops where faster generation can shorten the wait between tool calls.
- Visual UI Development: Interpreting screenshots and inspecting rendered interfaces to guide code revisions.
- Long-Document Assistants: Combining lengthy source material, charts, and repeated conversation context for analysis and structured extraction.
- Office Automation: Planning and revising presentations, reports, and spreadsheets through external authoring tools and visual checks.
Capabilities and Limitations
| Capability | Description |
|---|---|
| Reasoning | Always enabled; supports interleaved thinking during tool use and preserved thinking across turns. |
| Multimodal | Supports text, image, video, and file input and produces text output. |
| Context Window | 1,000,000 tokens. |
| Max Output | 131,072 tokens. |
| Tool Use | Supports function calling with optional streaming of tool-call arguments. |
| Structured Output | Supports JSON object output; strict JSON Schema enforcement is not documented as a model-specific guarantee. |
Known Limitations
- Thinking cannot be disabled. Selecting
lowreduces reasoning effort but does not create a non-thinking mode. - Preserved thinking requires complete, unmodified, correctly ordered historical
reasoning_content. Integrations must retain it when continuing the relevant reasoning and tool-call sequence.
Standard Pricing
The token prices below are shown in USD per 1 million tokens; web search is billed in USD per use.
| Model | Input (USD / 1M Tokens) | Cache Write (USD / 1M Tokens) | Cache Read (USD / 1M Tokens) | Output (USD / 1M Tokens) | Web Search (USD / use) |
|---|---|---|---|---|---|
| GLM-5.3-FlashX | $0.37 | $0.37 | $0.075 | $1.25 | - |
Credits settlement: B.AI converts charges at 1 USD = 1,000,000 Credits and deducts Credits from the account balance.
Prices shown in the documentation are B.AI standard reference prices for base billing purposes. B.AI may provide lower actual usage costs through limited-time offers, top-up bonuses, and account benefits. Specific prices, bonus Credits, account benefits, and final billing are subject to the platform display and billing records.