Hyperouter Newshttps://hyperouter.app/news/2026-11-06T00:00:00ZThe dashboard in five languageshttps://hyperouter.app/news/dashboard-in-five-languages2026-11-06T00:00:00Z2026-11-06T00:00:00ZHyperouterdash.hyperouter.app now shows in English, 简体中文, 日本語, Français and Español, with a new header.Seedream, Seedance, MiniMax H3 and Berryhttps://hyperouter.app/news/new-image-and-video-models2026-11-02T00:00:00Z2026-11-02T00:00:00ZHyperouterTwo image models and six video models join the catalog.Studio, a workstation for short dramashttps://hyperouter.app/news/studio-short-drama-workstation2026-10-31T00:00:00Z2026-10-31T00:00:00ZHyperouterStudio at studio.hyperouter.app turns a premise into a series of up to 64 episodes, on your own Hyperouter key.Low-credit alertshttps://hyperouter.app/news/low-credit-alerts2026-10-30T00:00:00Z2026-10-30T00:00:00ZHyperouterHyperouter tells you once when your balance falls below a threshold you set.GLM-5.3, GPT-6 and DeepSeek V4.1 Flash 1017https://hyperouter.app/news/new-chat-models-glm-gpt-6-deepseek2026-10-30T00:00:00Z2026-10-30T00:00:00ZHyperouterFour chat models join the catalog, called through the same /v1/chat/completions endpoint.Async images and video: respond-async, polling and webhookshttps://hyperouter.app/news/async-images-and-video-respond-async-polling-and-webhooks2026-10-26T00:00:00Z2026-10-26T00:00:00ZHyperouterHow to get an image or a video without holding a connection open. Send Prefer, get a 202, then poll GET /v1/generations/{id} or GET /v1/videos/{id}, or receive a webhook.Check your balance from code with GET /v1/creditshttps://hyperouter.app/news/check-your-balance-with-get-v1-credits2026-10-26T00:00:00Z2026-10-26T00:00:00ZHyperouterGET /v1/credits returns your balance, what open requests hold and what is left to spend, so a script can warn you before a request is refused.Errors you can act onhttps://hyperouter.app/news/errors-you-can-act-on2026-10-26T00:00:00Z2026-10-26T00:00:00ZHyperouterThe error object, the status codes worth handling, why Hyperouter answers 503 with X-Upstream-Status instead of 502 or 504, and a retry loop with backoff.Every request has an ID you can quotehttps://hyperouter.app/news/every-request-has-an-id-you-can-quote2026-10-26T00:00:00Z2026-10-26T00:00:00ZHyperouterChat, image and video answers carry an id. Look the request up with GET /v1/generations/{id}, and quote the id in a support ticket.Prompt expansion for image and videohttps://hyperouter.app/news/prompt-expansion-for-image-and-video2026-10-26T00:00:00Z2026-10-26T00:00:00ZHyperouterSet prompt_expand to true and Hyperouter rewrites your prompt for the model before it renders. The rewrite is billed on the same request, and a failed rewrite costs nothing.Retry safely with Idempotency-Keyhttps://hyperouter.app/news/retry-safely-with-idempotency-key2026-10-26T00:00:00Z2026-10-26T00:00:00ZHyperouterA timeout does not tell you whether Hyperouter got the request. Send an Idempotency-Key and a repeat starts no second job and costs nothing extra.Call any model with the OpenAI SDKhttps://hyperouter.app/news/call-any-model-with-the-openai-sdk2026-10-25T00:00:00Z2026-10-25T00:00:00ZHyperouterPoint the OpenAI SDK at Hyperouter's base URL, use your Hyperouter key, and call Grok and Qwen models with the code you already have.Hyperouter is generally availablehttps://hyperouter.app/news/hyperouter-is-live2026-10-25T00:00:00Z2026-10-25T00:00:00ZHyperouterOne OpenAI-compatible API and one prepaid balance for xAI's Grok and Qwen models, across chat, images and video, with DeepSeek coming next.