Embedding models, a new SDK and price drops
Hi vv,
It’s been a busy week, and we’ve been hard at work to bring you new ways to get the most out of AI models.
What’s new:
- OpenRouter now supports embedding models! Build your RAG pipelines, classification, search, and clustering apps without leaving the OpenRouter ecosystem.
- The OpenRouter TypeScript SDK is newly available in beta (with Python coming soon!):
- Auto-generated from OpenAPI specs - Easy discoverability of endpoints, parameters, and features via strong autocomplete in the IDE of your choice.
- Actionable error messages - Strong types and clear errors make it easy to debug your apps.
- Type-safe by default - Catch invalid configuration and API calls long before they reach production.
Price Drops:
Competition in the open-weight model ecosystem continues to drive prices down:
- Z.AI: GLM 4.6 (exacto) on DeepInfra - input token cost dropped by 25.0%.
- Qwen 3 VL 235B A22B Instruct on Fireworks - input and output token cost dropped by 26.7%.
Product updates:
- You can add a backup payment method for your auto-top-up to make sure you don’t run out of credits if the primary payment method fails. More improvements and auto-top-up alerts are coming soon.
- Simple grouping on the activity page lets you quickly visualize and summarize usage by model, API key, or key creator.
Best,
OpenRouter Team
For questions or comments, reach out to us on X or on Discord.
Click here to unsubscribe.

Comments
Post a Comment