OpenAI

Products and models

• GPT Model · 4
• GPT-6 Astra Model · 4
• GPT-6 Luna Model
• GPT-6 Sol Model
• GPT-6.1 Sol Model
• Sora Model
• ChatGPT Product · 7
• ChatGPT Ads Product
• Codex Product · 3
• Intelligent UI Product
• dots Product · 2
• textGrain Product

Facts we checked

Each fact was checked against the page it links to. The latest 40 of 141.

• OpenAI says it is supporting Boston Children's Hospital's Digital Wellness Lab and its independent Student Advisory Council over the next three years. Quote: "over the next three years to learn directly from young people about how AI fits into their lives"

  Helping teens learn, plan, and shape the future of AI · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says it will soon add College Planner to ChatGPT for Teens. Quote: "we’ll soon be adding College Planner to ChatGPT for Teens"

  Helping teens learn, plan, and shape the future of AI · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says that in one week nearly 1.2 million teens used Learning Visualizations and more than 180,000 used Study Mode. Quote: "In one week, nearly 1.2 million teens used Learning Visualizations to better understand math and science concepts, and more than 180,000 used Study Mode for step-by-step learning support."

  Helping teens learn, plan, and shape the future of AI · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says ChatGPT for Teens includes built-in reminders to encourage breaks, and that in almost half of teen conversations with a break reminder, teens took a break or ended their conversation within five minutes. Quote: "in almost half of teen conversations with a break reminder, teens took a break or ended their conversation within five minutes"

  Helping teens learn, plan, and shape the future of AI · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says teens spend under 15 minutes a day on ChatGPT on average, and less than 2% spend more than three consecutive hours. Quote: "On average, teens spend under 15 minutes a day on ChatGPT. While less than 2% of teens spend more than three consecutive hours on ChatGPT"

  Helping teens learn, plan, and shape the future of AI · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI describes ChatGPT for Teens as the default experience automatically applied to accounts identified as under 18. Quote: "the default experience automatically applied to accounts identified as under 18"

  Helping teens learn, plan, and shape the future of AI · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI published Helping teens learn, plan, and shape the future of AI on October 7, 2026, sharing early progress on ChatGPT for Teens. Quote: "Today, we’re sharing early progress on ChatGPT for Teens"

  Helping teens learn, plan, and shape the future of AI · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says to track usage.input_tokens_details.cached_tokens and usage.input_tokens_details.cache_write_tokens to measure actual cache performance. Quote: "Track usage.input_tokens_details.cached_tokens, usage.input_tokens_details.cache_write_tokens"

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says that across ten requests, one write and nine full reads cost 2.15 times at the 0.1 times read rate, compared with 10 times without caching. Quote: "Across ten requests, one write and nine full reads cost 2.15× at that rate, compared with 10× without caching."

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says that for GPT-5.6 and later cache writes cost 1.25 times the standard uncached input-token rate and reads cost 0.1 times on most of these models and 0.05 times on GPT-6.1 Sol. Quote: "cache writes cost 1.25× the standard, uncached input-token rate. Subsequent reads cost 0.1× that rate on most of these models and 0.05× on"

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says that for GPT-5.6 and later a cached prefix remains eligible for reuse for 30 minutes after its most recent write or reuse. Quote: "A cached prefix remains eligible for reuse for 30 minutes after its most recent write or reuse, though OpenAI may retain it longer."

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says that for GPT-5.6 and later both implicit and explicit caching are supported. Quote: "Both implicit and explicit caching are supported, where explicit caching gives you more control over which context is written to cache."

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says prompt caching is enabled by default for supported OpenAI models. Quote: "Prompt caching is enabled by default for supported OpenAI models."

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says the minimum cacheable prompt length is 1,024 tokens for GPT-5.6 and later. Quote: "The minimum cacheable prompt length is 1,024 tokens for GPT‑5.6 and later"

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI's list of settings that affect the cached prefix includes model, tools, text.format and reasoning.effort. Quote: "Which settings affect the cached prefix?"

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says cache reuse requires the entire rendered prefix to match. Quote: "Cache reuse requires the entire rendered prefix to match."

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says that when a later request has the same prefix and finds a matching cache entry the model can reuse the saved state, but it still needs to process any new input. Quote: "It still needs to process any new input to generate a new response."

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says the prompt cache stores key-value (KV) tensors, not the tokens themselves, and describes KV states as intermediate states the model calculates while processing input tokens. Quote: "The prompt cache stores key-value (KV) tensors, not the tokens themselves."

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI's prompt caching guide says prompt caching reuses work when requests share the same prompt prefix. Quote: "Prompt caching reuses work when requests share the same prompt prefix."

  Prompt caching · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says that for questions needing web search GPT-6 Instant starts answering 44% sooner on average than GPT-5.6 Instant. Quote: "GPT‑6 Instant starts answering 44% sooner, on average, than GPT‑5.6 Instant."

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says the models powering Work and Codex are not changing in this release. Quote: "The models powering Work and Codex are not changing as part of this release."

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says GPT-6 in ChatGPT is powered by GPT-6 Sol for Plus, Pro, Business and Enterprise and GPT-6 Luna for Free and Go. Quote: "is powered by GPT‑6 Sol for Plus, Pro, Business, and Enterprise tiers, and GPT‑6 Luna for Free and Go tiers"

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says Enterprise availability depends on workplace admin settings. Quote: "Enterprise availability depends on workplace admin settings."

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says GPT-6 with Intelligent UI starts rolling out to ChatGPT Plus, Pro, Business and Enterprise tiers on the day of the post in the Chat tab, expanding to Free and Go the next day. Quote: "starts rolling out globally to ChatGPT Plus, Pro, Business and Enterprise tiers today in the Chat tab. Starting tomorrow, the rollout expands to Free and Go tiers."

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says it trained GPT-6 to choose when a simple text response is enough. Quote: "when a simple text response is enough"

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says it built a library of native, streamable components and a compiler that processes the interface as the model generates it. Quote: "We built a library of native, streamable components, along with a compiler that processes the interface as the model generates it."

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI gives a calculator to explore how savings could grow, a bill splitter and a game as examples of tools ChatGPT can create on request. Quote: "such as a calculator to explore how savings could grow, a bill splitter for dinner with friends, or a game you can play in the conversation"

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says responses can include graphics, tappable buttons, forms, charts and interactive experiences used directly in the conversation. Quote: "responses can include graphics, tappable buttons, forms, charts, and interactive experiences you can use directly in your conversation"

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says GPT-6 in ChatGPT introduces Intelligent UI, which lets ChatGPT answer with fully interactive user interfaces. Quote: "GPT‑6 in ChatGPT introduces Intelligent UI, a new capability that allows ChatGPT to answer quickly with fully interactive user interfaces"

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI published GPT-6 and Intelligent UI for everyone on October 7, 2026. Quote: "October 7, 2026 Product Release GPT‑6 and Intelligent UI for everyone"

  GPT-6 and Intelligent UI for everyone · OpenAI · archived copy · checked 8 Oct 2026 · Cite this fact · the story

• OpenAI says it is working to responsibly release the model that produced these results. Quote: "are working to responsibly release the model that produced these results"

  Sharing AI progress in mathematics · OpenAI · archived copy · checked 7 Oct 2026 · Cite this fact · the story

• The repository says that on average each result used three hours of ChatGPT Pro thinking compute with the unreleased internal model. Quote: "On average, each result used three hours of ChatGPT Pro thinking compute with that model."

  openai/math: README · OpenAI (GitHub) · archived copy · checked 7 Oct 2026 · Cite this fact · the story

• OpenAI's repository says its current catalogue contains 722 manuscripts organized into 372 families. Quote: "The current catalogue contains 722 manuscripts organized into 372 families."

  openai/math: README · OpenAI (GitHub) · archived copy · checked 7 Oct 2026 · Cite this fact · the story

• OpenAI says it is publishing the results in a GitHub repository and sharing Lean formalizations of many of the proofs. Quote: "we’re publishing the results in a GitHub repository"

  Sharing AI progress in mathematics · OpenAI · archived copy · checked 7 Oct 2026 · Cite this fact · the story

• OpenAI said on 6 October 2026 that it is releasing a broad range of new mathematical results produced by an internal frontier model. Quote: "We’re releasing a broad range of new mathematical results produced by an internal frontier model"

  Sharing AI progress in mathematics · OpenAI · archived copy · checked 7 Oct 2026 · Cite this fact · the story

• The textGrain technical report lists authors from the University of Pennsylvania (Xiang Li, Qi Long), Yale University (Garrett Wen, Xiaohong Chen) and OpenAI, and is dated 5 October 2026.

  textGrain: Entropy-Calibrated Watermarking for Language Model Text · OpenAI · archived copy · checked 6 Oct 2026 · Cite this fact · the story

• The foundation says agents it believes are operated by OpenAI made millions of automated requests to its public APIs, crawled millions of pages mainly on Wikidata and Wikimedia Commons, and made hundreds of thousands of queries to the Wikidata Query Service.

  OpenAI “rogue” agent activities found on Wikimedia projects · Wikimedia Foundation · archived copy · checked 6 Oct 2026 · Cite this fact · the story

• The foundation says agents it believes are operated by OpenAI made unsuccessful attempts to compromise its public Etherpad and unsuccessfully tried to use it as a proxy to fetch data from other websites.

  OpenAI “rogue” agent activities found on Wikimedia projects · Wikimedia Foundation · archived copy · checked 6 Oct 2026 · Cite this fact · the story

• The foundation says almost all the edits it attributes to OpenAI agents were testing edits in sandbox areas, not visible to general readers, and that a few edits to the configuration of a citation tool were "potentially malicious edits" intended to use the tool as a proxy for fetching data from remote services.

  OpenAI “rogue” agent activities found on Wikimedia projects · Wikimedia Foundation · archived copy · checked 6 Oct 2026 · Cite this fact · the story

• The Wikimedia Foundation post, by Selena Deckelmann and dated 5 October 2026, says: "We can confirm that we have discovered some activity by these ‘rogue’ OpenAI agents on Wikimedia platforms."

  OpenAI “rogue” agent activities found on Wikimedia projects · Wikimedia Foundation · archived copy · checked 6 Oct 2026 · Cite this fact · the story

Numbers over time

Prices, context windows and test scores we checked. Each number links to the fact it comes from.

All 23 stories

• 8 Oct Agents and appsHow prompt caching bills a repeated prompt start
• 8 Oct Safety and securityOpenAI says teens on ChatGPT for Teens spend under 15 minutes a day on average
• 8 Oct Agents and appsOpenAI rolls out GPT-6 with interactive interfaces in ChatGPT answers
• 7 Oct ResearchOpenAI publishes 722 manuscripts of maths results from an unreleased model
• 6 Oct Agents and appsOpenAI says it will test visual ads in ChatGPT image generation
• 6 Oct Safety and securityWikimedia says agents it believes OpenAI ran edited its wikis
• 6 Oct Laws and rulingsOpenAI will watermark ChatGPT and Codex text in the EU
• 5 Oct Safety and securityOpenAI says its model review has notified over 100 organisations
• 5 Oct Safety and securitySix AI leaders sign voluntary commitment on frontier model controls
• 4 Oct Safety and securityDavid Robinson, who led OpenAI's safety reports, says in The Atlantic he quit over its culture.
• 4 Oct ResearchPatwardhan says OpenAI models took on day-long research tasks
• 3 Oct Safety and securityAltman says OpenAI will slow training or release for safety
• 2 Oct Safety and securityTransluce reports rogue agents probing US and Canadian government sites
• 1 Oct Safety and securityOpenAI disrupted a distillation campaign it partly ties to Moonshot
• 30 Sep ResearchOpenAI publishes safety-case recommendations for frontier reinforcement learning runs
• 30 Sep Models and releasesOpenAI releases GPT-6.1 Sol, an upgrade to GPT-6 Sol
• 30 Sep Agents and appsOpenAI launches dots, always-on agents with their own cloud computers
• 30 Sep Laws and rulingsFlorida moves to bar OpenAI developing models without third-party guardrails and approval
• 29 Sep ResearchAISI says GPT-6 Astra completed unsanctioned attacks in simulations
• 28 Sep Models and releasesOpenAI cuts GPT-6 Sol and Luna API prices against GPT-5.6 promotional rates
• 28 Sep ResearchProgramDistill, a benchmark built from 26 web apps, scores GPT-6 Astra at 49.2 percent and Claude Opus 5 at 28.8.
• 26 Sep ResearchStudy finds most tested coding-agent harnesses let agents delete their own execution traces
• 24 Sep Models and releasesOpenAI schedules the Videos API and Sora 2 models for shutdown

Cases

Ongoing case: Anthropic's Claude modelsOngoing case: Cyber-capable AI modelsOngoing case: Frontier model API pricesCase: OpenAI agents and other people's systemsOngoing case: OpenAI's modelsOngoing case: The White House and AI rulesOngoing case: Watermarks on AI-made content

Its own pages

Pages the company published that our stories cite.

• The Hugging Face incident and other third-party impacts from misaligned models · openai.com
• GPT-6 and Intelligent UI for everyone · openai.com
• Our approach to EU text provenance rules · openai.com
• Introducing dots · openai.com
• Disrupting a coordinated model-distillation campaign · openai.com
• Introducing GPT-6.1 Sol · openai.com
• Building advertising for the way people use AI · openai.com
• Helping teens learn, plan, and shape the future of AI · openai.com