AI Briefing
KOSign in
Pick

ox-alpha

·2026.08.21 21:40

Key point

OpenRouter has started offering the inference model 'Ox Alpha' from an anonymous provider for free.

Details

OpenRouter has launched a new stealth model, Ox Alpha. This model is an inference model for coding, persistent agent tasks, and production workloads, suitable for long-term software engineering, complex reasoning, and workflows combining text and visual context.

Ox Alpha is a model developed and operated by a third-party provider who remains anonymous, with OpenRouter only handling request routing. Prompts and completions are retained by the provider and are not used for training.

Key specifications and performance metrics are as follows:

  • Context window: 1,048,576 tokens
  • Max output: 131,072 tokens
  • Input/output price: Free
  • Throughput: 31 tps
  • Latency: 3.07s
  • Uptime: 100.00%

The model was released on August 20, 2026, and accepts text, image, and video inputs while returning text.

llmai-codingmodel-releaseopenrouterox-alphastealth-model
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

RedditAlibaba Releases Qwen 4 Max as 14-Day Preview ModelRedditQwen Head Confirms Qwen 4.0 and 5.0 Will Remain Open SourceHuggingFace BlogIBM Releases LLMON Markup Language and Granite 4.1 Preview Models with Enhanced Prompt Injection RobustnessPyTorchKRBotPsy-0.8B Achieves Higher Tool-Use Performance Than 2B ModelsTLDR AIStudy demonstrates LLMs can learn multi-step reasoning on filler tokensTLDR AIOpenRouter Analysis: MCP Consumes 42x More Context Tokens Than CLI for GitHub Tasks