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 Harness RC.8 adds native image-request configuration for DeepSeek adapters, image input for core commands, and on-demand Claude Code and Codex subagent bundles. The official release also improves search concurrency, Windows PowerShell sessions, storage, and the Python SDK.
Read articleSome DeepSeek users say they are seeing an earlier-looking reasoning style and better long-task results in V4 Pro. The reports may point to a traffic experiment, but DeepSeek has not confirmed a new model, checkpoint, or routing change.
Read articleDeepSeek will introduce separate peak and off-peak API pricing for V4 Flash and V4 Pro on August 17, 2026. This chart compares the old rates with the new prices for cache-hit input, cache-miss input, and output tokens.
Read articleDeepSeek has released its Harness developer preview under the MIT license. The agent runtime makes the model adapter, tools, session log, and even the agent loop replaceable through plugins.
Read articleDeepSeek V4 Pro 0813 went live at about 11 p.m. Beijing time on August 13. The model keeps the same API name and current rates, but DeepSeek is warning users to expect a significant price increase soon.
Read articleDeepSeek V4 Pro 0813 is now official. On August 17, DeepSeek will introduce peak and off-peak API pricing, with V4 Pro rates rising by as much as 12 times the previous price.
Read articleUpdated August 13: DeepSeek-V4-Pro-0813 is now live. This July 31 release introduced the official Flash build, its large agent gains, and native Responses API support for Codex.
Read articleUpdated August 13: DeepSeek V4 Pro 0813 is now live, largely confirming the reported mid-August model window. DeepSeek Harness remains pending.
Read articleUpdated August 13: DeepSeek V4 Pro 0813 is now live through the existing API model name. The earlier gray-rollout signals are now historical context; DeepSeek Harness remains pending.
Read articleMoonshot AI's Kimi K3 scores 57 on the current Artificial Analysis Intelligence Index, ahead of Claude Opus 4.8, GPT-5.5, and DeepSeek V4 Pro. Its full weights are promised for July 27, and its API price is 30% of Claude Fable 5's, not one eighth.
Read articleChecked on July 14, 2026: DeepSeek's official docs sitemap currently lists dedicated `chat_curl`, `chat_nodejs`, and `chat_python` sample pages, and all three use the current `https://api.deepseek.com` endpoint plus `deepseek-v4-pro` instead of older alias-first snippets.
Read articleChecked on July 13, 2026: DeepSeek's official docs sitemap currently lists both a top-level Prompt Library page and an Integrate with AI Tools guide covering Claude Code, OpenCode, and OpenClaw, giving developers a first-party docs path from prompt exploration to concrete tool configuration.
Read article