What changed on KoRouter.
Models added, price changes and gateway updates, newest first. Prices are in USD per 1M tokens as of each entry's date — the live price is always on the model page.
- Price
GLM-5.3 price cut: input $1.12 → $0.70 and output $3.52 → $2.20 per 1M tokens, 50% below Zhipu's official price.
- Price
deepseek-v4-pro and deepseek-v4-pro-0813 now cost the same as Flash: $0.024 input and $0.096 output per 1M tokens. Since 14 September DeepSeek serves every v4-pro request with V4.1-Flash and bills it at the Flash rate, so the official reference price shown for these models is the Flash price.
- New
deepseek-v4.1-flash is available — DeepSeek's current Flash model, at the same price as deepseek-v4-flash.
- Improved
GPT-6 Astra, Sol and Luna now bill requests above 272K input tokens at OpenAI's long-context rate (2× input and cache, 1.5× output), the same way OpenAI does. Each model page lists both tiers.
- Improved
Gateway core upgraded. Tool definitions whose schema has a null required list or tuple-style arrays no longer come back as a 400 from the upstream, and long streaming responses keep the connection alive while the model is still generating.
- Price
Free to sign up, no monthly fee — pay as you go with credits.