GLM 5.3 Flash Is Live — Native Multimodal, Budget Band
August 26, 2026
GLM 5.3 Flash is now in the Fast tab of the model picker. It is Z.ai's first native-multimodal GLM-5 model — text and images in one pass — on Z.ai's own API, at Budget 1×.
What you can do
- Pick GLM 5.3 Flash from the model selector — 1M-token context, image input, tools, Swarm worker eligibility, and always-on max-effort reasoning.
- Let Community Auto pick it. Flash is in the Auto pool for Community. Paid Auto skips it — quality 58 sits under the paid floor of 60, same pattern as Gemini 3.7 Flash.
- Keep the flagship separate. GLM 5.3 stays the Closed SOTA pick for long-horizon coding. Flash is the cheap sibling when you want the same family at Budget 1×.
- Pay Budget 1×. Z.ai is running a list-price promo through 9 September 2026; Alfrada bills the list band so your multiplier does not jump when the promo ends.
Where this shows up
You want GLM-family tool use and a 1M window without Premium 6×. Pin GLM 5.3 Flash instead of GLM 5.3.
You have screenshots, UI mocks, or slide renders in the thread. Flash can read the image and keep calling tools in the same turn.
You leave the picker on Auto on Community. Flash can now win everyday workhorse turns. On Pro, Pro Max, or Black, Auto still skips it and stays on models at quality 60 or above.
Try it
- [Pin GLM 5.3 Flash] "Read this screenshot and rebuild the layout as a responsive page, then check it in the browser."
- [Pin GLM 5.3 Flash] "Turn these notes and the attached chart image into a one-page memo with a risk list."
- [Pin GLM 5.3 Flash] "Triage this inbox of tickets, group them, and draft replies — keep it fast."
Heads up
- This is not GLM 5.3. The flagship stays on Premium 6×. Flash does not inherit those pins.
- Paid Auto does not pick Flash. Community Auto can. Pin it on a paid plan if you want that specific model.
- Calls go to Z.ai (
api.z.ai), not OpenRouter. This is not a ZDR pin. - Thinking cannot be disabled. Alfrada always sends max-effort reasoning, same as GLM 5.3.