New coverage from across AI, software, and hardware.
On 13 August 2026 xAI expanded github.com/xai-org/x-algorithm with ranking weights, visibility-filtering code, and Phoenix training code — Apache 2.0, not a dump by accident. What is public, what is withheld, and what actually changes for a For You user.
Cursor has been acquired by SpaceX, completing an April partnership and gaining access to the largest GPU fleet for model training. Grok 4.6 launches as the first joint model, matching GPT-5.6 Sol on key benchmarks and available now in Cursor.
On 10 August 2026 OpenAI put GPT-5.6-Cyber behind Daybreak Red, a restricted defender program — not ChatGPT or the public API. Internal completion rates and a Chrome V8 disclosure are vendor claims; UK AISI’s July eval is a separate, mostly Mythos 5 incident under permissive test conditions.
IBM and OpenAI announced a strategic partnership embedding GPT-5.6, Codex, and ChatGPT Work into IBM Consulting Advantage. The deal creates a dedicated OpenAI practice with thousands of certified consultants and expands cybersecurity collaboration through IBM Autonomous Security.
OpenAI replaced CRO Denise Dresser after nine months, hiring Wiz president Dali Rajic to lead sales. The shake-up follows other executive departures as the company prepares for a potential IPO.
Z.ai launched GLM-5.3 on 14 August 2026 on the same base as GLM-5.2, claiming large coding and cyber gains from post-training alone. Open weights are due in about two weeks; Coding Plan users already get the new model.
DeepSeek, Qwen, and GLM-class APIs often price an order of magnitude below Claude Opus, GPT-5.6 Sol, and peers. The useful question is which tier matches the job you intend — and how close the outcome gets to that intention without paying for unused ceiling.
A recent shadow evaluation covered by Nature found an AI agent using Claude Opus 4.8 failed at open-ended research judgment despite strong engineering execution. Anthropic’s own RSI essay confirms an 8x code acceleration but admits the same judgment gap persists.
OpenAI opened a preview of Ultrafast, an API tier that runs GPT-5.6 Sol up to 14 times faster on Cerebras wafer-scale chips. The mode targets enterprise workloads like incident response and financial analysis with up to 750 output tokens per second.
