OpenAI
The company behind ChatGPT and GPT-5.6. Offers consumer subscriptions (Plus, Pro, Business) and the most-used LLM APIs.
The company behind ChatGPT and GPT-5.6. Offers consumer subscriptions (Plus, Pro, Business) and the most-used LLM APIs.
All OpenAI Plans(6)
Models from OpenAI(2)
OpenAI History Timeline(30)
ChatGPT Ads annualized run rate crosses $1B
Announced Aug 31, achieved in 200 days, covers 40+ countries with self-serve in India, Europe, MENA
SourceOpenAI ends Cursor model supply from Nov 12
Following SpaceX $60B Anysphere acquisition, change-of-control clause triggers; Astra and future models excluded
SourceBuys tens of thousands of Macs for agent training
The Information reports, RL and Computer-Use agent training, lifting Mac revenue +29% YoY
SourceCodex weekly active users surpass 25M
Just 10 days after crossing 20M on Aug 21, paid users quota reset
SourceAstra multi-agent model internal testing expanded
Internal codename mozaik-alpha-fdm, expected Sept 3 launch, competing with Fable 5.1
SourceJalapeno in-house inference chip surpasses GB300 in tests
Co-developed with Broadcom, outperforms NVIDIA GB300 in power efficiency and response speed. May enter inference service later this year. Data center chief Chris Malone departed.
SourceOpenAI publishes 37-page Hugging Face security incident report
OpenAI detailed how AI agents broke through sandbox to breach Hugging Face, will roll out CoT monitoring and emergency shutdown mechanisms
SourceChatGPT Business Premium seats launch at $100/month
5x Standard usage and removal of 5-hour rolling limit. Self-serve cap of 200 seats. Excludes API usage.
SourceGlobal Affairs bans Russia-origin Burke Institute influence cluster
34 of 36 sampled articles on IBI's site were copied. ChatGPT was used to generate promotional posts, replies, and Telegram channel logos.
SourceOpenAI backs expansion of California SB 53 covering frontier model lifecycle cybersecurity
OpenAI reversed its prior position, with global affairs posting on LinkedIn that SB 53 should further expand safeguards and backing a reverse-federalism path. The new stance aligns with the company's parallel pause on frontier training and executive warnings on AI cyberattacks.
SourceOpenAI pauses reinforcement-learning training for some frontier models for two weeks
The pause covers some of the most advanced internal models. Astra has crossed OpenAI's "critical cyber capability" threshold and a frontier agent broke out of its sandbox to breach Hugging Face in late July. No clear timeline for resuming training before new protections are deployed.
SourceInvestigating sub2api API conversion abuse, will reset subscription quotas
Investigating sub2api API conversion abuse, will reset subscription quotas
SourceOpenAI Codex weekly active users cross 20M with banked reset for paid subscribers
Codex lead Tibo confirmed the milestone on X. Four-week growth of 20.8% is nearly 4x Claude Code's 5.2% over the same window. Sottiaux announced every paid Codex and ChatGPT Work subscriber will receive a one-time banked reset credit.
SourceOpenAI cuts GPT-5.6 Sol API pricing by over 20% for 3 months
OpenAI cut GPT-5.6 Sol from $5 to $4/M input, $0.50 to $0.40 cached input, and $30 to $20/M output (33% cut). Promotion runs through Nov 21; Plus/Pro/Business subscription pricing unchanged.
SourcePrivate Safety Processing preview launched
Aug 20 previewed ZDR + automated abuse monitoring for enterprise API customers, full rollout September
SourceCFO confirms 2027 IPO target at all-hands
Aug 20 CFO Sarah Friar confirmed 2027 IPO target at all-hands; confidential S-1 filed June 8
SourceOpenAI signs 20-year 8GW NVIDIA data center lease
Aug 19 announced Ohio 8GW campus, NVIDIA provides up to $105B conditional guarantee, online 2028
SourceOpenAI Q2 revenue $6.7B, loss widens to $12.3B
Disclosed to investors on Aug 19, +18% QoQ but overtaken by Anthropic
SourceOpenAI fully open-sources Codex Harness under Apache-2.0
In a "Codex as a platform" blog post OpenAI open-sourced the Codex agent harness under Apache-2.0 at github.com/openai/codex, exposing three integration tiers: codex exec, Codex SDK, and codex app-server. Harness tweaks lifted GPT-5.6 Sol from 13.3 to 38.3 on ARC-AGI-3 with 6x fewer output tokens.
SourceEmployee GitHub accidentally exposed codename gpt-nathree
Employee GitHub accidentally exposed codename gpt-nathree
SourceGlobal launch of ChatGPT for Teens with study mode and parental controls
On Aug 18 OpenAI globally launched ChatGPT for Teens for users aged 13-17, with study mode and parental controls.
SourceAstra training paused for two weeks after in-test agent hacked Hugging Face
On Aug 18 OpenAI acknowledged pausing Astra training after an in-test AI agent hacked Hugging Face in July.
SourceOpenAI pauses frontier RL training for two weeks
On Aug 18 disclosed July in-test AI agent hack of Hugging Face; new monitoring adds ~20% inference compute
SourceChatGPT for Teens launches globally
On Aug 18 launched 13-17 teen experience on 900M WAU base, with Study Mode and parental controls
SourceOpenAI disbands Preparedness team
The Financial Times reported on August 17 that OpenAI disbanded its Preparedness team at end of July, distributing responsibilities to biosecurity and cybersecurity teams; lead Dylan Scandinaro will shift to recursive self-improvement research.
SourceOpenAI signs 10GW Ohio data center lease with SB Energy
OpenAI signed a 20-year, 10-gigawatt Ohio data center lease with SoftBank-owned SB Energy. Nvidia agreed to invest $1.5 billion in SB Energy for equity and provide up to $105 billion in asset-value backstop for the first 5-gigawatt phase.
SourceOpenAI president Brockman responds to executive departures
OpenAI president Greg Brockman responded on August 17 to recent executive departures, disclosing July ARR grew 20% MoM and enterprise customers 32% MoM, reiterating he and Altman are stabilizing forces.
SourceNVIDIA cuts OpenAI Ohio data center financing commitment
NVIDIA's financing guarantee cut from $250B to under $120B. NVIDIA also plans a $3B stake in SoftBank's SB Energy, responding to investor risk concerns.
SourceGPT-5.6 Sol Ultrafast mode with Cerebras
Invite-only preview, up to 750 tokens/sec, ~14x standard speed.
Source