GPT-5.6 Luna goes free with unlimited chats

OpenAI removed the text-chat ceiling for free ChatGPT users and shipped model upgrades on both tiers — the day's biggest AI story, with Cloudflare, Suno, and a startup funding round close behind.

OpenAI is making GPT-5.6 Luna the default model for Free and Go users, with unlimited text chats, as ChatGPT passes 1 billion weekly users. Luna replaces GPT-5.5 and adds a "Think" button for higher-reasoning requests. Plus and Pro users get an upgraded GPT-5.6 Sol tuned for quick tasks — a separate model from the Codex/Work version — plus a thinking slider. OpenAI's internal evals claim 62% fewer factual errors for Luna and 68% fewer for Sol versus GPT-5.5-Instant. Sol is live today; the free-tier changes roll out this week and next, and limits on files, images, and voice stay.

The take: killing caps on the free tier is how you defend a billion weekly users against a wave of free rivals, and the Sol/Luna split signals that OpenAI now treats ChatGPT as a multi-model product, not a single model with a price tag.

Cloudflare open-sourced Cloudflare OS, the internal platform that lets non-coders build apps by describing workflows to AI agents. Thousands of employees use it daily; each app instance runs in a V8 isolate rather than a container — about 100x faster to start and 10–100x more memory-efficient — with agents starting at zero permissions and outbound networking off by default. One catch: the backend only deploys on Workers Paid plans, a requirement Cloudflare initially buried and has since flagged up front after GitHub complaints. This is the most credible answer yet to the "vibe coding is a security nightmare" objection — sandboxing as the default, not an afterthought.

Suno is adding watermarking, fingerprinting, and download limits to fight the flood of spammy AI tracks. CEO Mikey Shulman's new post promises transparency tools aligned with "emerging industry standards," including EU AI Act labeling rules, plus downloads restricted to paying subscribers with monthly caps, following last year's Warner Music settlement. Specifics are still thin, and Shulman stops short of judging AI-made art, leaving disclosure to artists and platforms. Watermarks that streaming services actually honor are the only thing that will keep AI music from burying human releases — good on Suno for moving first, but the proof will be in the distribution deals.

Naïve raised $28.5 million to let AI agents handle the drudgery of starting and running a company. The startup's API packages payments, email, phone numbers, cloud infrastructure, storage, and US LLC incorporation behind a single endpoint that tools like Cursor, Claude Code, and Codex can drive, and it says 30,000+ developers signed up within months of launch. Vibe coding has moved from "build the app" to "build the business" — the agents do the incorporation, you supply the token budget. The obvious question: what happens to the people who used to sell those services?

What to watch: the full Free/Go rollout of GPT-5.6 Luna next week — and whether unlimited free chats pushes Google and Anthropic to loosen their own free tiers.

Is unlimited free access to a frontier model sustainable — or just a land grab for users? Tell us in the comments.

Akhetonics says its all-optical CPU reaches a customer in 2026

Light-based computing keeps promising more than it delivers — but one Munich startup has just put a date on its bet, and the interview laying it out is doing the rounds on Hacker News this week. Akhetonics says it will deploy its first commercial machine with a major customer by the end of 2026, with several more planned for 2027. The company, founded by Michael Kissner and Leonardo Del Bino, is building a computer where data enters as light, is switched as light, and circulates through memory

The Week in AI — October 5–11, 2026

Every big claim this week turned out to rest on fine print more interesting than the headline: revenue only the company reporting it can define, safety tests sandboxed while the product keeps the web, and a Pentagon phase-out nobody would confirm until reporters kept asking. The week's top 5 1. OpenAI's revenue was $20 billion below the numbers everyone quoted — and the gap was definitional. The Financial Times reported Thursday that OpenAI's annualized revenue runs roughly $20 billion unde

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Russia's largest tech company is learning what the AI era's infrastructure war looks like from the receiving end — three data centers in four days, and with them much of the cloud layer Russian businesses run on. A Ukrainian drone strike knocked out Yandex's data center in Vladimir early Sunday morning, the third of the company's facilities hit since October 8. The site — reported at roughly 50 MW and designed for about 2,880 server racks — stopped operating completely after the attack, Yandex

Agent teams cost up to 5x more, barely score higher

The multi-agent hype train hit a benchmark this weekend — and the grid and the trucking regulators had quiet weeks of their own. Vals AI put agent teams head-to-head with single agents on its Vibe Code Bench, and the teams cost between 1.8 and 5.1 times more for almost no extra quality. The evals company ran GPT-6 Sol and Claude Opus 5.5 solo and in teams across 50 apps at two reasoning efforts; out of four comparisons, only one was statistically significant — Sol at medium effort, where the t

Read more

Akhetonics says its all-optical CPU reaches a customer in 2026

The Week in AI — October 5–11, 2026

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Agent teams cost up to 5x more, barely score higher