Archive
GPT-6 Astra, Claude Fable 5.1, and Gemini 3.8 Live are all live and shipping at the same time. Here's the full September 2026 benchmark round-up — who leads on reasoning, coding, math, multilingual, voice, and cost, and what the gaps actually mean for picking a model.
Contact AIPress for tips, corrections, speaking inquiries, and reader feedback. We read every message and correct errors promptly.
AIPress privacy policy: how we collect and protect visitor information, our use of cookies and tracking, third-party links, and data security practices.
AIPress is a daily AI news and analysis publication covering AGI, ChatGPT, Claude, OpenAI, Anthropic, AI infrastructure, safety, and DeepSeek — original reporting from primary sources, published every day.
Google's Gemini 3.8 Live Extended Thinking took #1 on Artificial Analysis' Speech-to-Speech Quality Index at 82.6 — available today via the Gemini API for developers building voice agents.
In her 2026 State of the Union, Ursula von der Leyen invited frontier AI labs to Brussels to 'pace the frontier' — and endorsed the slowdown call even as President Trump called AI doomsday warnings a 'HOAX.'
DeepSeek engineer Liu Shengyu likened Anthropic getting AGI to 'Hitler with the A-bomb' in a WeChat post that has become the sharpest Chinese-industry rebuttal to Amodei's pacing call — and moved markets across two continents.
Anthropic's September 2026 threat intelligence report is the most detailed disclosure yet from a frontier lab on how its own models are being weaponized — Russian state espionage automating malware evasion, a Bangladeshi operative running 29 Claude accounts for 16 months, and named Chinese labs stealing Claude's reasoning traces 300,000 times in ten days.
Nvidia CEO Jensen Huang declared AGI has arrived on X, Sept 7, pointing at GPT-6 Astra trained on 100K+ Grace Blackwell systems. Two days later at Goldman Sachs he forecast 70% revenue growth for 2027 — possibly over 100%. Here's what he actually said, what his stack looks like, and why a chip CEO's AGI call is different from a model lab's.
Claude Fable 5.1 ships September 1 — leads Opus 5 on every benchmark Anthropic published, 45% cheaper for agentic workloads at the same sticker price. Plus Mythos 5.1, the science pitch, and what breaks for Fable 5 users migrating up.
DevsIsle designs and ships AI systems, agents and integrations for teams that need it done properly.
