Very cheap per token and multimodal, which makes it a sensible default for budget agents and office automation.
Kenning's take
What it does
GLM-5.3-Flash is the first GLM-5 model that takes images, video and files as input. It has 320B total parameters with 18B active, a 1M-token context, and is meant for cheap coding and office work such as slides and spreadsheets. It was tested anonymously on OpenRouter before launch and is available via the Z.ai API, the GLM Coding Plan and MIT-licensed open weights, with a faster FlashX version.
- Video, image, text and file input
- Checks rendered output visually and revises it
- 1M-token context
- 320B total, 18B active parameters
- Faster GLM-5.3-FlashX option
Who it suits. Developers who want a cheap model for visual coding, agents and document work.
Key facts
- Category
- AI models
- Part of
- GLM-5.3
- Made by
- Z.ai (Zhipu AI)
- Based in
- China
- Released
- 26 August 2026
- How to get it
- Z.ai API (models glm-5.3-flash and glm-5.3-flashx), GLM Coding Plan, or open weights on Hugging Face
- Works on
- API, open weights (Hugging Face)
- In the UK
- Yes
- API price
- $0.15 per million input tokens, $0.50 per million output (FlashX $0.37 and $1.25)
Pricing
| Plan | Price | Billed |
|---|---|---|
| GLM Coding Plan | $from 18 | month |
FAQ
When was GLM-5.3-Flash released?
Z.ai listed it in its release notes on 26 August 2026.
How much does GLM-5.3-Flash cost?
$0.15 per million input tokens and $0.50 per million output on the Z.ai API.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.
Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.
Cisco Antares
Cisco's small open-weight models that find where known security vulnerabilities sit in a codebase, running locally.