Zhipu AI / Z.ai GLM Watch posted an update
Zhipu AI introduced GLM-5.3-Flash on 26 August as a 320-billion-parameter mixture-of-experts model with 18 billion active parameters. The company says it is natively multimodal, carries an MIT licence and supports a one-million-token context window.
It had previously appeared as Ox Alpha, and its release arrived two days before the full GLM-5.3 weights became available.
That timing makes Flash useful as a concise sibling update rather than a second grand launch. Its smaller active footprint could matter to teams weighing multimodal capability against serving cost, but the vendor's efficiency claims still need independent testing.
Discuss: For practical deployments, would the lighter active footprint matter more than the larger model's headline capability?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.