OpenAI says it is partnering with Google Cloud to put its frontier models and Codex agentic workflows in Google Cloud's Model Garden, with general availability planned in coming weeks. Customers can buy them through the Google Cloud Marketplace and apply GCP commitments. Microsoft remains OpenAI's primary cloud partner.
OpenAI says its Patch the Planet coalition coordinated fixes for critical PHP vulnerabilities that Calif found with GPT-5.6 Sol and that Wiz independently reported. PHP maintainers led the patches. OpenAI committed $1 billion in subsidized Daybreak access for Frontline Defenders.
EAGLE-4 publishes inference code and eagle-harness, which adds its speculative decoding to SGLang, vLLM and TokenSpeed. The EAGLE-4 Team claims up to 3.42x faster generation on Qwen3.8-27B with four B300 GPUs. The source-available license permits commercial use below $20 million in aggregate revenue.
Radical Numerics says phages designed by its Omnii-1 genome language model have been given to a patient with an antibiotic-resistant infection under a physician-led compassionate-use pathway, which the company calls the first clinical use of AI-generated phages. The company reports the cocktail cleared the patient's isolate in vitro but did not disclose outcomes.
The NSA warns that two critical PHP flaws affect roughly 70% of websites and says unpatched sites should go offline; patches exist and the current version is PHP 8.5.12. OpenAI says a Calif researcher using GPT-5.6 Sol found one hole, and Wiz says its AI agents found others. OpenAI says no exploits have been reported.
Christian Keil argues in an a16z essay that AI models still cannot carry a hardware design from CAD file to assembled part, because translating bits into atoms and back stays lossy. He cites tests of OpenAI's Astra, Anthropic's Fable 5.1 and Grok 4.6 on his LuxoBench lamp project. SpaceX lists 591 supply chain employees.
Alphabet CEO Sundar Pichai says Google's cloud unit forgoes short-term revenue to steer compute toward frontier AI research, calling that allocation non-negotiable. He says Gemini 4 Argon processed nearly 25 trillion tokens in its first 72 hours, about three times Gemini 3, and that Google remains capacity constrained serving it.
METR reviews Anthropic's August 2026 risk report and agrees misalignment risk from covered models, including Mythos 5, is low, but says that reflects limited capabilities rather than alignment training or monitoring. METR's Nikola Jurkovic, Tim Hua and Beth Barnes say Anthropic's covert-capability benchmarks poorly test evasion of its real monitoring system, and cite incidents that undercut the report's argument.
Mistral releases Voxtral 3, a family of speech-to-text and text-to-speech models plus a voice agent SDK, claiming state-of-the-art multilingual transcription across 50 or more languages. The company prices Voxtral Mini Transcribe 3 at $0.0035 per minute in batch and Voxtral Mini TTS 3 at $13 per million characters. It runs on Mistral's EU-hosted API or on customer infrastructure.
Meta releases Muse Spark Open, an open-weight mixture-of-experts model with about 816 billion parameters, 42 billion active per token, and a 1 million token context window. It accepts text, images, audio and video and generates text, aimed at agentic coding. Meta calls it the most capable US open-weights model. The BF16 checkpoint is about 1.5 TB.