why aren't all 129 resolved? a statement only gets an assessment when the public
record can support or contradict it. opinions and what-ifs never can, and 0 checkable
ones are still open, waiting for their date. predictions held up or didn't;
assertions are supported or contradicted. on every card:
▮▮▮▮▮ certainty ·
▮▮▮▮▮ debate potential. speakers are clickable
Assertion Contradicted
Sonwalkar: GPT-5 costs half as much as o3
“Also it's half the cost of O three. So it's much cheaper. So it helps you, helps your margins.”
Assertion Supported
Apple requested custom AI models from OpenAI and Anthropic for Private Cloud
“In his discussions with both Anthropic and OpenAI, the iPhone maker requested a custom version that could run on Apple's private cloud computer.”
Assertion Supported
Chen: OpenAI reached 3 million paying business users
“We hit a big milestone. We got I think three million paying business users fairly recently.”
Assertion Supported
LiveKit's d'Sa: LiveKit built the infrastructure for OpenAI's ChatGPT voice modes
“Fast forward, like, about a year and a half, two years, and, ah, we built a demo when ChatGPT first came out, ah, where you could talk to it instead of text with it using LiveKit's infrastructure, and a few months later, OpenAI ended up finding it, ah, wanted …”
Assertion Supported
D'Sa: LiveKit partnered with OpenAI to build ChatGPT voice mode
“We built a demo when ChatGPT first came out, ah, where you could talk to it instead of text with it using LiveKit's infrastructure, and a few months later, OpenAI ended up finding it, ah, wanted to build a voice interface to ChatGPT, And we started to work pre…”
Assertion Supported
Coogan: OpenAI's o1 Pro beats R1 and Claude in Catan eval
“The top OpenAI thinking models, O-one Pro get it too, but all of DeepSeek R-one, Gemini, two point O-flash thinking and Claude do not.”
Assertion Supported
Coogan: OpenAI's $200 o1 Pro uses the same model as $20 o1
“In fact, the O-one model is 20 dollars a month, is basically the same model used in the O-one Pro model for 10 X the price at 200 dollars a month, which raised plenty of eyebrows. The main difference is that O-one Pro thinks for a lot longer before responding”
Assertion Supported
Coogan: Stargate's First Texas Data Center to Be Run by Oracle
“Stargate's first data center will be in Texas. The site which started construction last year will be operated by Oracle and used by OpenAI.”
Assertion Partly supported
Coogan: OpenAI Mac app leverages accessibility features to edit code across editors
“They use the accessibility features on the Mac app so that they don't need VS codes or text edits permission to go in there and rewrite the code. They just are able to read everything out, rewrite it, and just act as a AI assistant in there.”
Assertion Contradicted
Coogan: OpenAI o1 is built on the GPT-4 foundation model
“The foundation model, yeah, it's still running GPT-IV, but obviously there's a lot of there's a lot of training to have that happens on top and a lot of UI stuff.”
Assertion Supported
Hays: OpenAI Built Consumer Scale After Wrappers Reached Tens of Millions ARR
“Before chat GPT, there were products like copy.ai that were leveraging the API well, and like had gotten to decent, not scale, but tens of millions of ARR, and then OpenAI was like, oh, actually we're a consumer product company now, and then now they have hund…”
Assertion Supported
Coogan: GPT-6 Astra scores zero percent on Exploit Gym Honeypot
“Also, the Exploit Jim Honeypot Lower is the better. GPT-Six Astra, zero percent, so good performance there.”
Assertion Supported
Coogan: GPT-6 Astra scores 64.6% on Terminal Bench Science
“Terminal Bench Science. We're going over into the world of science. GPT-Six Astra scored 64.6% on reasoning effort. Max cost 26, 26 bucks.”
Assertion Supported
OpenAI Acquired Jony Ive and Tang Tan's AI Startup in 2025
“Tang Tan left Apple in 2024 to co-found the AI hardware startup I.O. With Johnny Ive, which OpenAI acquired in 2025 as part of its consumer AI device effort.”
Assertion Supported
Hays: OpenAI GPT-5.6 spans Sol, Terra, and Luna tiers
“To go back to your other question, so Sol is the frontier model, Terra is a balanced model for efficient everyday work, and then Luna is a fast and affordable model for high volume work.”
Assertion Supported
OpenAI introduces Codex Sites for building shareable interactive web apps
“In other open AI news there's now sites in codecs. You can turn work ideas, plans into interactive websites or apps. Your team can explore, use, and share.”
Assertion Supported
Hays: ChatGPT launches personal finance features
“ChatGPT launched personal finance features.”
Assertion Supported
OpenAI Codex prompt allegedly bans mentioning goblins, raccoons, and pigeons.
“Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures. Unless it is absolutely and unambiguously relevant to the user's query.”
Assertion Supported
OpenAI's GPT-5.5 matches GPT-5.4 per-token latency in real-world serving
“GPT 5.5 delivers this step up in intelligence without compromising on speed. Matches GPT 5.4 per token latency in real world serving.”
Assertion Supported
OpenAI hires Meta's Dave Dugan as VP of Global Ad Solutions
“Dave Dugan, a former top advertising executive at Meta, is gonna lead sale, ad sales. Dugan, who announced earlier this month he was stepping down from his role as Vice President of Global Clients and Agencies at Meta, has been named Vice President of Global A…”
Assertion Supported
McCormick: OpenAI signed a deal with Cerebras for fast inference
“OpenAI just did the Cerebrus deal, there's Claude Fast, and there's a whole bunch of different modes that will deliver faster inference”
Assertion Supported
Coogan: OpenAI launched GPT-5.3 Codex and Frontier AI co-worker platform
“So GPT-Five-Point-Three Codex launched today, and there's a very, very cool There's a very, very cool update to or new product, Frontier, which is a product for building AI co-workers that I'm very excited to talk to Sam about because it feels like the first g…”
Assertion Supported
Coogan: 20 OpenAI Employees Left to Join Thinking Machines at Inception
“20 people from OpenAI left during the Thinking Machines Foundation.”
Assertion Supported
Coogan: GPT-5.2 Thinking Benchmark Leaps to 52.9% from 17.6%
“GPT 5.2 thinking is at 52.9%, up from 17.6%. That's a huge leap.”
Assertion Supported
Hays: OpenAI's Sora Fell from #1 to #21 in App Store Charts
“Sora was, I think, number one in the charts at that point. It's now 21.”
Assertion Partly supported
Nadella: Azure was OpenAI's first cloud provider after Musk requested credits
“Azure was the first cloud provider. When OpenAI got started. In fact, I think Elon sent me The mail asking for Azure credits.”
Assertion Supported
Hays: OpenAI announces web browser named Atlas
“OpenAI dropped a trailer announcing the browser, which is called Atlas.”
Assertion Supported
Hays: 85% of ChatGPT's weekly active users are outside the US
“85% of their weekly actives are outside of the U.S.”
Assertion Supported
Alex Dhillon: OpenAI is a customer of Outtake AI
“We've been lucky to work with phenomenal places like you know, OpenAI is one of our favorite customers really pushed us when you think about the kinds of cybersecurity threats that they might be dealing with.”
Assertion Supported
Knoop: OpenAI asked ARC Prize team to verify results on semi-private dataset
“Opening eye situation was a little different because they had reached out to us and said, Hey, we think we've got a really impressive result on the public eval set. And we'd like your help to verify. On the semi-private set, which is what we created that data …”
Assertion Supported
Coogan: OpenAI Will Run Its Own Cluster Management Software for Stargate
“OpenAI will main control of operations and run almost all of its own cluster management software.”
Prediction Held up
Coogan: AI apps will dynamically route user prompts across models automatically
“Very quickly, they're going to be able to put a router in front of that and dynamically decide, oh, this is a question that needs to go to oh, one. This is a question that needs to go to four. Four O or whatever, and all of that will become extracted.”
Assertion Supported
Isaac: OpenAI-Musk Trial Courtroom Limits Outlets to One Reserved Seat
“So there are reserved seats in the front row for press, but the issue is only one person per outlet gets it, and we are doing, like, live blogging for, like, the big moments, like, opening statements, and for Elon, and so myself and my colleague Cade came on t…”