Vercel and OpenRouter cut GLM-5.3 API pricing for a limited window
Vercel and OpenRouter are offering limited-time 50% pricing windows for Z.ai's GLM-5.3 family, with different model slugs, providers, and end times.
Two AI gateways are running time-limited 50% pricing windows for Z.ai's GLM-5.3 family. Vercel's AI Gateway promotion covers zai/glm-5.3-promo-50 through Tuesday, September 8, while OpenRouter lists a separate discount for GLM-5.3-Flash through September 9, 2026 at 16:00 UTC.
The offers are useful for short evaluation cycles, coding-agent pilots, and batch experiments. They are separate promotions with different model slugs and routing rules, so copying one provider's example into another gateway will not preserve the discount.
Vercel's GLM-5.3 promotion is provider-pinned
Vercel says the promotional slug zai/glm-5.3-promo-50 routes only through DigitalOcean. It does not fall back to another provider, and Vercel says the slug stops serving when the offer ends. The standard zai/glm-5.3 slug remains available after September 8 at normal rates; provider options can prefer DigitalOcean while retaining fallback behavior.
The model accepts text input, exposes a 1M-token context window, and supports up to 128K output tokens according to Vercel's announcement. Vercel also documents setup paths for Claude Code, Codex, OpenCode, Cursor, and other coding-agent clients.
OpenRouter lists a separate GLM-5.3-Flash window
OpenRouter's current model listing identifies GLM-5.3-Flash as a native multimodal Z.ai model for efficient coding and long-horizon agent tasks. The listing shows a 50% discount via ZAI until September 9 at 16:00 UTC, alongside a 1,310,720-token context window and published per-million-token pricing.
OpenRouter routes across multiple providers and displays uptime and performance information. That makes its offer operationally different from Vercel's DigitalOcean-only promotional slug. Treat the model name, provider selection, and expiration time as configuration data in any test harness.
What developers should verify before using the offers
- Confirm the exact promotional model ID in the gateway dashboard before sending production traffic.
- Record the end time in UTC and add a fallback to a standard model slug before the promotion expires.
- Check whether provider pinning changes latency, regional availability, data handling, or reliability for your workload.
- Measure total cost with prompt caching, retries, tool calls, and long context included; the headline discount applies to the listed offer, not every downstream cost.
- Keep evaluation results separate from normal-rate runs so the temporary price does not distort your cost comparison.
For agent routing patterns and fallback design, see LinkLoot's AI workflow automation guide.
Sources and timing
Vercel is the primary source for the DigitalOcean AI Gateway promotion, its promotional slug, routing behavior, and September 8 end date. OpenRouter independently corroborates a live 50% GLM-5.3-family discount while documenting a separate Flash model, provider network, pricing, and September 9 at 16:00 UTC deadline. Neither offer should be treated as a permanent price cut.
Try the related loot
Put six hosted Workers AI models behind Cloudflare AI Search
