Sign InOpen Brain
VercelEngineering PostOfficial Source

GLM-5.3 is 50% off through DigitalOcean on AI Gateway

GLM-5.3 is half-price through September 8 via a temporary DigitalOcean-only model ID. Keep the standard ID in durable agent configs if you need fallback after the offer.

Vercel · Sep 2, 2026
Open Source Open MarkdownOpen JSON
Source Summary

Vercel is offering **50% off GLM-5.3 through September 8** using zai/glm-5.3-promo-50. That identifier routes only to DigitalOcean, has no provider fallback, and stops serving when the offer ends.

Practical Implication

Treat the promo ID as temporary configuration rather than committed application logic. For continuity, keep **zai/glm-5.3** and prefer DigitalOcean through the gateway provider order; other providers can then take over at their normal rates.

Agent-Ready Context
Vercel is offering **50% off GLM-5.3 through September 8** using zai/glm-5.3-promo-50. That identifier routes only to DigitalOcean, has no provider fallback, and stops serving when the offer ends.

Treat the promo ID as temporary configuration rather than committed application logic. For continuity, keep **zai/glm-5.3** and prefer DigitalOcean through the gateway provider order; other providers can then take over at their normal rates.

The model accepts text with a **1M-token context window** and up to **128K output tokens**. The material gives no discounted token prices, and savings disappear when requests fall back from DigitalOcean or the promotion expires.
Connected Context · Feed7 Judgment

This sharpens the distinction between a promotional route and a durable model choice. Unlike discounts that preserve the normal model ID, this offer requires a provider-locked ID that expires completely; continuity therefore depends on keeping the standard GLM-5.3 route and expressing DigitalOcean as a temporary provider preference. The large context window expands possible trials but does not establish lasting cost or quality advantages.

Context Map
infracoding#gateways#coding-agents
Uncertainty
The model accepts text with a **1M-token context window** and up to **128K output tokens**. The material gives no discounted token prices, and savings disappear when requests fall back from DigitalOcean or the promotion expires.