OpenAI
RuntimeWire's continuing coverage of OpenAI — GPT-5.6 and the model roadmap, ChatGPT as a platform, Codex, safety disclosures, and the business behind the scaling bet.
OpenAI is the company that turned large language models into a consumer product — ChatGPT — and then into an economy. RuntimeWire covers it the way a wire covers a central bank: not because every announcement matters, but because enough of them move everyone else's roadmap.
Our OpenAI coverage runs from the GPT-5.6 model family (Sol, Luna, and the speed-tiered variants) and the Codex coding stack to ChatGPT's push into payments, desktop computing, and the enterprise — plus the harder stories: cyber-capability evaluations, an Astra release that keeps slipping on safety reviews, and the financial engineering (employee tender offers, infrastructure commitments, IPO speculation) behind the scaling bet.
What we're tracking:
- Model releases and what they actually change for builders — pricing, speed, context, capability
- ChatGPT as a platform: payments, agents, skills, and distribution deals
- Safety and security: red-team disclosures, capability pauses, and what OpenAI publishes about its own failures
- The money: compute deals, secondaries, and the path to an eventual IPO
Every story links to primary sources — model cards, filings, and first-party posts — so you can check our work.
All updates
- OpenAI benchmarks a 16x load-time gain for giant ChatGPT and Codex threads — A 741-turn, 231 MB thread loaded in 1.66 seconds in a test, with lower memory growth and 98% fewer requests.
- OpenAI built a credit bonus into ChatGPT’s experimental desktop onboarding — A hidden flow asks newcomers to let ChatGPT complete a real desktop task, then can add a remotely configured bonus to their balance.
- Anthropic stole ChatGPT's aura. Then it gave it back. — Claude Code made Anthropic Silicon Valley's chosen winner. Its models and business remain formidable.
- ChatGPT Wallet: OpenAI Is Building Agentic Payments — Code in the Windows ChatGPT/Codex desktop application bundle shows secure card entry and states for saving payment methods that ChatGPT can use to complete tasks.
- OpenAI launches ChatGPT Computer History to remember work across Mac apps — The opt-in feature turns screen activity into reusable context for Pro, Business, and Enterprise users, with a timeline and privacy controls.
- OpenAI acquired 4.2% of Cerebras before GPT-5.6 Ultrafast launch — RuntimeWire found that OpenAI acquired a 4.2% economic stake in Cerebras weeks before launching Cerebras-powered GPT‑5.6 Ultrafast.
- OpenAI previews GPT-5.6 Sol at up to 14x standard speed — Cerebras hardware drives the API tier to 750 output tokens per second, with access restricted to selected customers while capacity grows.
- OpenAI data finds top enterprise users adopt reusable AI skills six times as often — OpenAI's customer telemetry links heavier use to Plugins, reusable skills and Codex, offering startup operators a deployment playbook without proving those tools improve results.
- OpenAI starts rolling out giftable ChatGPT credits — The one-time claim links let users buy prepaid usage for someone else. OpenAI's Windows app already contains gated entry points for the purchase flow.
- OpenAI launches ChatGPT for Linux, bringing Codex to major desktop distributions — The preview supports Ubuntu, Debian and Fedora, closing a platform gap after Anthropic shipped Claude for Linux in July.
- OpenAI, Anthropic and Google blocked a cross-model reasoning attack — The attack used weaker sibling models as decoders and exposed credentials in public agent logs before providers blocked it.
- OpenAI buys $7 billion of employee shares before a possible IPO — OpenAI funded the liquidity itself, preserving its $852 billion price while turning private stock into cash for current and former staff.
- OpenAI pledges to pay its way after Texas freezes data center approvals — The AI lab promised new power support, efficient cooling and usage disclosures as Texas audits projects seeking grid connections.
- OpenAI’s GPT-5.6-Cyber answers 95% of exploit requests, up from 1.5% for Sol — Daybreak Red limits the purpose-trained model to approved researchers, while Daybreak Blue opens GPT-5.6 Sol for broader defensive work.
- OpenAI pauses some Astra work as it assesses critical cyber capabilities — OpenAI's unreleased model showed strong cyber performance, prompting isolated testing and universal monitoring while OpenAI continues its assessment.
- OpenAI's Sam Altman said the singularity had already started — The OpenAI CEO used a July 25th podcast to define the singularity as a gradual shift centered on access, control and compounding AI research.
- OpenAI acquires NextSlide as ChatGPT expands into PowerPoint — Founder Ahmed Beshry is selling a second startup after Caper, and his new team will work on ChatGPT's creation and communication tools.
- OpenAI slows Astra release after cyber tests raise critical-risk concerns — Sam Altman says OpenAI still plans broad access, but the lab is tightening controls around a model with potentially critical cyber capabilities.
- Black Hat publishes full OpenAI presentation on Hugging Face agent breach — The upload confirms RuntimeWire's earlier reporting from closed captions obtained through a QR code leaked from Greg Brockman's personal X account.
- Former OpenAI researcher Gabriel Petersson launches Energy for AI-driven computer work — The Sora 2 contributor is betting that a desktop interface can bring coding-agent workflows to the rest of knowledge work.
- OpenAI makes GPT-5.6 Luna the default for free ChatGPT users — Free and Go users will get unlimited everyday chats and a Think button, while Plus and Pro users get an updated Sol model with adjustable reasoning.
- OpenAI joins Amazon, Microsoft and Cursor on portable agent plugin standard — The working draft packages Agent Skills and MCP servers so one plugin can run across competing AI clients, with Vercel also helping govern the specification.
- OpenAI brings APA into its youth mental health safeguards — The agreement turns APA from an outside reviewer into a partner on resources for families, clinicians and schools.
- Unverified leak points to an OpenAI Astra launch next week — The account behind the claim calls Astra OpenAI's largest pretrain since GPT-4.5 and names "mewfour" as its release candidate.
- OpenAI agents rebuilt a secret message board after the company shut it down — OpenAI resumed training two days after its models took over an internal server and caused an outage.
- Castform and Neon say a 4B model matched GPT-5.6 Sol at 100x lower cost — A Neon case study says the specialized open-weight model matched GPT-5.6 Sol on retrieval accuracy while costing about 100x less per inference workload.
- Naomi Bashkansky left OpenAI for Conduit's telepathy bet — The former alignment researcher joined a San Francisco team training models on thousands of hours of non-invasive neural recordings.
- OpenAI discloses two cyber evaluations where models reached real systems — GPT-5.6 Sol used public tokens and tunneling services in one test; an unnamed model exploited a real website in another.
- OpenAI hires reverse-engineering veteran Halvar Flake for cyber and efficiency work — The incoming OpenAI researcher previously founded security startup zynamics and profiling-software maker Optimyze, selling both companies.
- OpenAI releases messages challenging Apple's hardware trade-secret claims — The records challenge parts of Apple's case while raising new questions about files moved during an engineer's departure.