DeepSeek V4: Million-Token Context That Actually Works

<p>At the International AI Summit in Dublin, OpenAI Chief Financial Officer Sarah Friar took the stage to announce that the lab is holding back a breakthrough model. According to

<p><a class="article-body-image-wrapper"

<p>In the first eight weeks of 2026, ten major open-weight LLM architectures were released.</p> <p>GLM-5 matched GPT-5.2 and Claude Opus 4.6 on benchmarks. Step 3.5 Flash

<p>The price page hasn't moved. Five dollars per million input. Twenty-five per million output. Identical to Opus 4.6.</p> <p>Run the same Claude Code call on both, though, and

<p>DeepSeek shipped V4-Flash-0731 last week — same 284B parameter architecture as the preview, same 13B activated parameters per token, MIT licensed, open weights on HuggingFace.

<p>The week of July 11 to 18, 2026 delivered news at every layer of the AI stack. Moonshot AI shipped the largest open-weight model ever announced, Google targeted its

<a href="https://news.ycombinator.com/item?id=47885014">Comments</a>

<ul> <li> <strong>Book:</strong> <a href="https://www.amazon.com/dp/B0GX2SQ594" rel="noopener noreferrer">System Design Pocket Guide: Interviews</a> </li> <li> <strong>Also by