Follow the latest DeepSeek model releases, API changes, pricing and developer tools, alongside news from the wider AI ecosystem. DeepSeek V4 Hub is an independent publication.
DeepSeek's official status page records a web and API incident starting at 08:11 Beijing time on October 15. The V4 Pro API, V4.1 Flash API and chat service were affected. At the 12:33 Beijing time check, DeepSeek had identified the cause and was implementing a fix; the incident was not marked resolved. There is no official evidence linking the outage to a V4.1 Pro gray test.
Read articleMeta's Muse Wonder 1.0 Standard API costs $4 per million input tokens, $20 per million output tokens, and $0.20 per million cached input tokens. DeepSeek V4.1 Flash lists lower token rates, with separate peak and off-peak prices. Muse Wonder Full has no published Model API token rate for this comparison. Prices checked October 14, 2026.
Read articleGoogle opened Gemini 4 Argon to paid API users on October 13. Its Standard launch rates are $2 input, $10 output, and $0.10 cached input per million tokens through November 30; DeepSeek V4.1 Flash lists lower off-peak and peak token rates. Prices checked October 14, 2026.
Read articleGLM-5.3 Highspeed has appeared in ZCode, with an early user reporting 490 output tokens per second. Pi lists the model as glm-5.3-highspeed. ZCode now documents a randomized GLM HighSpeed trial for subscribers; the reported speed and the proposed Omen Alpha connection remain unverified.
Read articleA developer going by FeiZ has open-sourced DeepSeek Bot, a community project built on DeepSeek Harness. It is available as a standalone desktop app or a DSH plugin, with named teammates, models, and memory.
Read articleDeepSeek's DSec infrastructure supports stateful agent training with four sandbox backends, composable environments, on-demand image loading, and rollout recovery. Read our technical analysis and a full English translation of DeepSeek's Chinese article.
Read articleDeepSeek Harness v0.2 is now in preview, with desktop installers for macOS and Windows, built-in tools for everyday work and development, and a new plugin management page.
Read articleDeepSeek will cut V4 Flash off-peak API rates at 12:00 Beijing time on September 10, lowering cache-hit input from RMB 0.05 to 0.02 per million tokens, cache-miss input from RMB 1.50 to 1.00, and output from RMB 4.50 to 4.00. Peak rates remain twice the off-peak price.
Read articleA temporary DeepSeek V4.1 Flash route is available for limited testing through the existing API base URL. The reported beta keeps V4 Flash billing, caps each account at 20 concurrent requests, and uses a model ID that points to a September 10 expiry.
Read articleDeepSeek Harness v0.1.2-alpha.3 improves long-session navigation and rendering, fixes queued image delivery, and removes the optional SQLite Session backend. The alpha remains separate from npm's default RC release.
Read articleA hands-on test asks DeepSeek, ChatGPT, and Gemini to recognize the same image and turn Girl with a Pearl Earring into a Perler bead pattern, with the DeepSeek and GPT results shown side by side.
Read articleDeepSeek has launched the experimental V4 Flash Vision Exp model on its API, with mixed image and text input, three request formats, and a free Files API for reusable uploads.
Read article