| Source | Rumors cited | Confirmed | Debunked | Pending | Hit rate |
|---|---|---|---|---|---|
| RuntimeWire | 2 | 0 | 0 | 1 | — |
| Startup Fortune | 2 | 1 | 0 | 1 | 100% |
| AGI Hunt — Multiple leaks point to OpenAI internal model GPT-6.2, codename Betelgeuse | 1 | 0 | 0 | 1 | — |
| AI Weekly | 1 | 0 | 0 | 1 | — |
| AIToolsRecap (daily roundups) | 1 | 1 | 0 | 0 | 100% |
| AIstify — single-source caution | 1 | 0 | 0 | 1 | — |
| APIMODELS LLM API catalog | 1 | 1 | 0 | 0 | 100% |
| APIMaster — Gemini 4 Krypton Preview: checkpoint leak, Carbon comparison and release expectations (checked Nov 16) | 1 | 0 | 0 | 1 | — |
| AdTechSumo — Meta's Muse Agent Begins Using Signed IDs in Limited Test, DataDome Reports | 1 | 0 | 0 | 1 | — |
| Azure Foundry region availability (no gpt-6-astra-minor listed) | 1 | 1 | 0 | 0 | 100% |
| Azure Foundry region availability (no gpt-6-sol listed) | 1 | 1 | 0 | 0 | 100% |
| BeInCrypto — OpenAI and Anthropic Weighed Mutual AI Tests | 1 | 0 | 0 | 1 | — |
| BeatAPI: What Is Space Bunny? | 1 | 1 | 0 | 0 | 100% |
| Bitsminds (DevDay recap) | 1 | 0 | 0 | 1 | — |
| CXO Digital Pulse — OpenAI Cuts Funding Target by $10 Billion, Seeks $20 Billion at $1.25 Trillion Valuation | 1 | 0 | 0 | 1 | — |
| CellCog | 1 | 0 | 0 | 1 | — |
| CellCog — ChatGPT Family Plan: leaked prices from $8 to $500 | 1 | 0 | 0 | 1 | — |
| CellCog — Fable 5.5 Release Date and the Hidden Obsidian 3 (Oct 28 update) | 1 | 0 | 0 | 1 | — |
| Cryptika — Critical ChatGPT GitHub connector flaw report | 1 | 0 | 0 | 1 | — |
| Crypto Briefing | 1 | 1 | 0 | 0 | 100% |
| CryptoBriefing — OpenAI quietly lists three aurora alpha models in its API | 1 | 0 | 0 | 1 | — |
| Cyber Security News — Critical ChatGPT GitHub connector flaw could allow attackers to steal private repository secrets | 1 | 0 | 0 | 1 | — |
| Daily Agentic — Anthropic's Mythos page briefly named Claude Mythos 5.5, then removed it | 1 | 0 | 0 | 1 | — |
| Dealroom — OpenAI cuts its funding round target by $10B to about $20B as the AI sell-off bites | 1 | 0 | 0 | 1 | — |
| DevXplore (Medium) — Anthropic and Nscale's $90 billion AI compute race | 1 | 0 | 0 | 1 | — |
| Digit — OpenAI reportedly wants to bring back Sora AI, but not everyone may be able to use it | 1 | 0 | 0 | 1 | — |
| EvoLink | 1 | 0 | 1 | 0 | 0% |
| ExplainX — AI News Today (FrontierCode 2.0 / Obsidian 3 claim checked) | 1 | 0 | 0 | 1 | — |
| Finimize — OpenAI lines up a $30 billion bridge round | 1 | 0 | 0 | 1 | — |
| First Hackers News — ChatGPT GitHub integration flaw: risks to private code | 1 | 0 | 0 | 1 | — |
| ForkLog — OpenAI considers transferring Sora to Hollywood startup | 1 | 0 | 0 | 1 | — |
| Forkast — GPT-6 Cyber would be the fourth cybersecurity model in twelve months | 1 | 0 | 1 | 0 | 0% |
| Geeky Gadgets | 1 | 0 | 0 | 1 | — |
| KWSN (Reuters) — China's DeepSeek, MiniMax win approval for AI global use | 1 | 0 | 0 | 1 | — |
| Kingy AI — GPT-6.2 Bel Rumors: Betelgeuse, Day 28 and the Evidence | 1 | 0 | 0 | 1 | — |
| LLM Rumors (Sol launch speculation) | 1 | 1 | 0 | 0 | 100% |
| Lapaas Voice — Meta Muse Spark 1.4 leak reveals unreleased model tested internally (Nov 14) | 1 | 0 | 0 | 1 | — |
| Lapaas Voice — OpenAI and Anthropic boost Google TPU orders, extending demand through 2029 (Nov 16) | 1 | 0 | 0 | 1 | — |
| Latent Space AINews — Gemini 4 Carbon tops leaderboards (FrontierCode 2.0 mystery model section) | 1 | 0 | 0 | 1 | — |
| LinkLoot (Altman delay) | 1 | 1 | 0 | 0 | 100% |
| LiteLLM Models — Claude Fable 5.5 | 1 | 0 | 0 | 1 | — |
| LiteLLM PR #46201 — add azure_ai/claude-fable-5-5 | 1 | 0 | 0 | 1 | — |
| Longbridge (CNBC report) — OpenAI nears up to $30 billion funding round at $1.25 trillion valuation led by UAE investors | 1 | 0 | 0 | 1 | — |
| MadRobot — Claude Fable 5.5 release date: is it out yet? | 1 | 0 | 0 | 1 | — |
| Madhyamam — OpenAI, Anthropic near agreement to cross-test AI models for security risks | 1 | 0 | 0 | 1 | — |
| NILE1 | 1 | 0 | 1 | 0 | 0% |
| OpenAI Developer Community (subscriber report) | 1 | 1 | 0 | 0 | 100% |
| OpenAI Developer Community — GPT-6.1 Sol suddenly consuming much more Plus usage since October 16 | 1 | 0 | 0 | 1 | — |
| OpenAI Help Center (Daybreak troubleshooting — names Astra Minor) | 1 | 1 | 0 | 0 | 100% |
| OpenAI Help Center (Daybreak trusted access — lists Astra Minor) | 1 | 1 | 0 | 0 | 100% |
| OpenAI community forum (Sol tier reports) | 1 | 1 | 0 | 0 | 100% |
| OpenRouter (GPT Sol Latest page \u2014 names the Pro slugs) | 1 | 1 | 0 | 0 | 100% |
| OpenRouter GPT-6 Migration Guide (officially documents the -pro variants) | 1 | 1 | 0 | 0 | 100% |
| PPC Land — Meta's Muse agent gains a signed ID in a quiet test, DataDome finds | 1 | 0 | 0 | 1 | — |
| PYMNTS | 1 | 0 | 1 | 0 | 0% |
| PYMNTS (citing The Information) | 1 | 1 | 0 | 0 | 100% |
| PYMNTS — Trump administration weighs naming and shaming US firms using Chinese AI (Nov 18) | 1 | 0 | 0 | 1 | — |
| Pagegoo (Fortune) | 1 | 0 | 1 | 0 | 0% |
| Pasquale Pillitteri — OpenAI Reportedly Finished Bel, the Model After GPT-6 (updated Nov 1 with the GPT-6.2 Astra sighting) | 1 | 0 | 0 | 1 | — |
| Pasquale Pillitteri — OpenAI confirms Codex limits error on 40,000 accounts but doesn't mention Sol | 1 | 0 | 0 | 1 | — |
| Reuters | 1 | 0 | 0 | 1 | — |
| Reuters (via WNCY) | 1 | 0 | 1 | 0 | 0% |
| Reuters via AmeriLedger — Trump administration weighs forcing US firms to disclose Chinese AI use (Nov 18) | 1 | 0 | 0 | 1 | — |
| Reuters — China's DeepSeek, MiniMax win approval for AI global use, state media-affiliated entity says | 1 | 0 | 0 | 1 | — |
| Reuters — Meta to price Muse Charm gadget at $300, The Information reports (Nov 19) | 1 | 0 | 0 | 1 | — |
| Reuters — Nscale in talks on $45 billion Anthropic cloud deal, Bloomberg News reports | 1 | 0 | 0 | 1 | — |
| RobinsPost (via Android Authority) | 1 | 1 | 0 | 0 | 100% |
| RuntimeWire — Anthropic's code hints at a gated way to CC Claude on email | 1 | 0 | 0 | 1 | — |
| RuntimeWire — Fable 5.5 appears in Azure metadata, not its model catalog | 1 | 0 | 0 | 1 | — |
| RuntimeWire — Three new Aurora model names pop up in OpenAI's API | 1 | 0 | 0 | 1 | — |
| Seeking Alpha — Meta sets $300 price on pocket-sized Muse Charm AI device: report (Nov 19) | 1 | 0 | 0 | 1 | — |
| Seoul Economic Daily | 1 | 0 | 0 | 1 | — |
| SignalPlus | 1 | 0 | 0 | 1 | — |
| SiliconANGLE — Gemini 4 Argon reaches Gemini app subscribers alongside a new tasks mode | 1 | 1 | 0 | 0 | 100% |
| StartupFortune (MiniMax link) | 1 | 1 | 0 | 0 | 100% |
| Stocktwits — Top supply chain analyst says new OpenAI, Anthropic chip orders boost Google's TPU outlook (Nov 16) | 1 | 0 | 0 | 1 | — |
| TIKR | 1 | 0 | 0 | 1 | — |
| TechCityAuthority — Claude Fable 5.5 reaches Anthropic partners, launch possible next week | 1 | 0 | 0 | 1 | — |
| TechCrunch | 1 | 0 | 0 | 1 | — |
| TechCrunch — Meta's new Muse Charm device may cost $300 (Nov 19) | 1 | 0 | 0 | 1 | — |
| TechCrunch — OpenAI working on new $30B raise at $1.25T valuation ahead of IPO | 1 | 0 | 0 | 1 | — |
| TechDefused — Bel vs Astra | 1 | 0 | 0 | 1 | — |
| TechGig | 1 | 0 | 1 | 0 | 0% |
| TechTimes — Treasury weighs Chinese AI disclosure labels for US firms (Nov 19) | 1 | 0 | 0 | 1 | — |
| Testing Catalog (Tuesday tease) | 1 | 1 | 0 | 0 | 100% |
| TestingCatalog — Anthropic prepares to release Claude Fable 5.5 next week | 1 | 0 | 0 | 1 | — |
| TestingCatalog — Anthropic prepares to release Claude Mythos 5.5 model (Nov 1) | 1 | 0 | 0 | 1 | — |
| TestingCatalog — Meta Muse Spark 1.4 surfaces in RocksDB pull request (Nov 13) | 1 | 0 | 0 | 1 | — |
| TestingCatalog — Meta prepares Muse Confidential VM for private AI work | 1 | 0 | 0 | 1 | — |
| TestingCatalog — OpenAI testing new 'Aurora' GPT-Image model on Arena | 1 | 0 | 0 | 1 | — |
| TestingCatalog — OpenAI to announce “o” always-on agent during DevDay (includes Sottiaux reset post) | 1 | 1 | 0 | 0 | 100% |
| The Big Ant — Nscale S-1/A: second-building option exercised Sept 29, 60 days to agree terms (Nov 15) | 1 | 0 | 0 | 1 | — |
| The Daily Fledgling — Morning Edition, November 8, 2026 (quoting Wang on X) | 1 | 0 | 0 | 1 | — |
| The Decoder — OpenAI lowers valuation target, raises up to $30 billion | 1 | 0 | 0 | 1 | — |
| The Hindu BusinessLine — OpenAI funding target cut by $10 billion as AI slumps | 1 | 0 | 0 | 1 | — |
| ThursdAI (Sept 19 episode) | 1 | 1 | 0 | 0 | 100% |
| Tibo Reset — Tibo's October 20 post on the 40,000-account limits inconsistency (quoted verbatim) | 1 | 0 | 0 | 1 | — |
| TradingKey — OpenAI and Anthropic Near AI Model Safety Cross-Testing Agreement | 1 | 0 | 0 | 1 | — |
| Unite.AI — Meta's Muse Charm price and free Verizon 5G plan detailed in report (Nov 19) | 1 | 0 | 0 | 1 | — |
| Unite.AI — OpenAI reopens ChatGPT Pro $200 sign-ups with new usage calculation | 1 | 1 | 0 | 0 | 100% |
| Van Data Team — GPT-6.2 Astra and Fable 5.5 leaks: what's real so far | 1 | 0 | 0 | 1 | — |
| WCCFTech — Bel's 10T parameters | 1 | 0 | 0 | 1 | — |
| Wccftech — Claude Fable 5.5 tipped to land this week | 1 | 0 | 0 | 1 | — |
| Wccftech — OpenAI Teases GPT-6.1 Astra Amid Relentless Hype Around The Mysterious GPT-6.2 Betelgeuse Model | 1 | 0 | 0 | 1 | — |
| WinCentral | 1 | 1 | 0 | 0 | 100% |
| World of AI (YouTube) | 1 | 1 | 0 | 0 | 100% |
| Zubiqo — Anthropic website listing hints at unannounced Claude Mythos 5.5 model (Nov 1) | 1 | 0 | 0 | 1 | — |
| Zubiqo — OpenAI quietly lists three undocumented Aurora alpha models in developer API | 1 | 0 | 0 | 1 | — |
| aiidelist | 1 | 1 | 0 | 0 | 100% |
| cellcog.ai (context: the Sept identifier-leak pattern) | 1 | 1 | 0 | 0 | 100% |
| cellcog.ai tracker | 1 | 1 | 0 | 0 | 100% |
| edgen.tech | 1 | 1 | 0 | 0 | 100% |
| edgen.tech (API rerouting leak) | 1 | 1 | 0 | 0 | 100% |
| explainx.ai | 1 | 0 | 1 | 0 | 0% |
| lupAI — OpenAI plans to revive Sora as a professional film tool | 1 | 0 | 0 | 1 | — |
| pricepertoken.com New Models Today tracker | 1 | 1 | 0 | 0 | 100% |
| spacebunny.im builder's guide | 1 | 1 | 0 | 0 | 100% |
| tokenmaxxing OpenRouter leaderboard | 1 | 1 | 0 | 0 | 100% |
Hit rate = confirmed ÷ (confirmed + debunked). Only rumor entries count — announced and released items aren't predictions. Early days: scores get meaningful as rumors resolve.
← Back to headlines