Archive

DeepSeek closed out the week by dropping V4 in two preview models — Pro and Flash — reclaiming the open-weights crown from Kimi and GLM and undercutting every Western frontier on price. OpenAI shipped GPT-5.5 as the next step toward its "super app," Google Cloud Next unveiled two new TPU generations and the Gemini Enterprise Agent Platform, and Anthropic locked in multiple gigawatts of TPU compute through a Broadcom-and-Google deal.

Anthropic dominated the week: Bloomberg's long-awaited feature revealed why Claude Mythos was deemed too dangerous to release (the model autonomously discovered and exploited an FreeBSD RCE during eval), while the company announced a $50B commitment to American AI infrastructure and attracted investor offers at an $800B valuation. Claude Opus 4.7 landed with stronger coding and vision at unchanged pricing.

Anthropic announced Claude Mythos 5 (10 trillion parameters) exclusively via Glasswing, a 12-partner cybersecurity initiative launching this week. Meanwhile, Anthropic surpassed OpenAI in revenue ($30B vs $25B), and the three frontier labs formed a coalition against Chinese model theft.

OpenAI closed a record $122B round at an $852B valuation, cementing the compute-first strategy that now defines the race. Cursor 3 reimagined the IDE as a multi-agent workspace, Alibaba shipped Qwen3.5-Omni with native text-image-audio-video processing, and Cognichip raised $60M to let AI design the chips that power AI.

ARC Prize dropped ARC-AGI-3 this week and every frontier model scored under 1% — humans still hit 100%. Anthropic is eyeing an October IPO at $60B+, OpenAI tacked another $10B onto its record fundraise, and Claude shipped Auto Mode for autonomous task execution.

NVIDIA dominated the week with GTC 2026 — Nemotron 3 Super is a 120B open model that runs at 12B efficiency, and NemoClaw gives agents a sandboxed runtime. OpenAI shipped GPT-5.4 Mini and Nano for high-volume workloads, Google launched Stitch as a full AI-native design canvas, Mistral unveiled Forge for enterprise model training, and Anthropic made the 1M context window generally available.

Claude turned security researcher this week, discovering 22 Firefox vulnerabilities including 14 high-severity flaws — the kind of result that reframes what frontier models can do beyond coding. Cursor is reportedly raising at a $50B valuation, OpenAI acquired red-teaming startup Promptfoo, Meta unveiled four custom AI chips, Variant launched a scroll-first AI design tool, and Perplexity shipped a persistent Mac mini agent called Personal Computer.

The Pentagon designated Anthropic a "supply chain risk" this week, sending shockwaves through the AI industry and pushing Claude to number one on the App Store as consumers rallied behind the company's refusal to build autonomous weapons. OpenAI closed a staggering $110B round at a $730B valuation and launched GPT-5.4, Cursor crossed $2B in annualized revenue, Google shipped Gemini 3.1 Flash-Lite, and Luma debuted creative AI agents powered by unified multimodal intelligence.