# GLM-5.3 Flash

> Z.ai multimodal model with a 1M-token context

- Category: Agents & dev (`agents`)
- Official: https://docs.z.ai/guides/vlm/glm-5.3-flash
- Wunderlust: https://wunderlust.xyz/r/glm-5-3-flash
- Image: https://wunderlust.xyz/media/glm-5-3-flash-poster.jpg
- Video preview: https://wunderlust.xyz/media/glm-5-3-flash.webm
- Added: 2026-11-05

## What it is

GLM-5.3-Flash and FlashX are Z.ai models, the first natively multimodal ones in the GLM-5 series: they accept video, image, text and files and respond in text.

## What it’s for

For evaluating a multimodal model with function calling, structured JSON output and long context.

## Benefits

- 1M-token context and output of up to 128K
- Function calling, streaming, context caching and structured JSON
- Understands video and images

## Caveats

- According to the maker, FlashX reaches about 200 tokens/s and the benchmarks are favorable; the figures are Z.ai's.
- The page doesn't state per-token API pricing.

## Similar on Wunderlust

- [Qwen3.8-Max](https://wunderlust.xyz/r/qwen3-8-max.md): Repository for Qwen3.8-Max, Alibaba Cloud's MoE model
- [Gemma 4](https://wunderlust.xyz/r/gemma-4.md): Open Gemma 4 model evaluated in local AI tests on Macs
- [Poppy AI](https://wunderlust.xyz/r/poppy-ai.md): Chat panel with context template boards for writing in the user's voice
- [Relay (Obsidian plugin)](https://wunderlust.xyz/r/relay-obsidian.md): Obsidian plugin for sharing a vault's context with agents
- [OpenViking](https://wunderlust.xyz/r/openviking.md): Context database for agents that organizes memory and skills like a file system
- [Buzzy](https://wunderlust.xyz/r/buzzy.md): AI platform for video and images, with Skills and an MCP server