OpenAI Releases GPT-6.2 Astra AI Model With Lower Prices and Stronger Safeguards Hot: Odds moved 13 points in 24 hours
OpenAI launches its advanced GPT-6.2 Astra model, reducing API costs by 20% while demonstrating improved capabilities in chip design and robotics applications.
Headline and summary written by an AI model from the linked articles; not reviewed by a person. How this was made
The odds
Each tick is a source publishing on this story
Price data
| Time | Chance of yes |
|---|---|
| now | 1% |
| 21 Nov, 09:19 UTC | 1% |
| 21 Nov, 04:18 UTC | 1% |
| 20 Nov, 23:19 UTC | 1% |
| 20 Nov, 22:21 UTC | 1% |
| 20 Nov, 14:17 UTC | 1% |
| 20 Nov, 13:15 UTC | 1% |
| 20 Nov, 03:17 UTC | 1% |
| 19 Nov, 22:54 UTC | 1% |
| 19 Nov, 15:40 UTC | 1% |
| 19 Nov, 15:19 UTC | 1% |
| 19 Nov, 11:24 UTC | 1% |
| 19 Nov, 08:42 UTC | 1% |
| 19 Nov, 06:40 UTC | 1% |
| 19 Nov, 05:34 UTC | 1% |
| 19 Nov, 04:26 UTC | 1% |
| 19 Nov, 03:59 UTC | 1% |
| 19 Nov, 00:03 UTC | 1% |
| 18 Nov, 20:16 UTC | 1% |
| 18 Nov, 17:32 UTC | 2% |
| 18 Nov, 13:04 UTC | 2% |
| 18 Nov, 12:57 UTC | 2% |
| 18 Nov, 11:13 UTC | 2% |
| 18 Nov, 10:44 UTC | 2% |
| 18 Nov, 10:24 UTC | 2% |
| 18 Nov, 09:41 UTC | 2% |
| 18 Nov, 09:15 UTC | 2% |
| 18 Nov, 08:42 UTC | 2% |
| 18 Nov, 08:29 UTC | 2% |
| 18 Nov, 03:40 UTC | 1% |
| 18 Nov, 03:00 UTC | 2% |
| 18 Nov, 02:53 UTC | 2% |
| 18 Nov, 01:25 UTC | 2% |
| 18 Nov, 00:58 UTC | 2% |
| 17 Nov, 23:30 UTC | 2% |
| 17 Nov, 23:10 UTC | 2% |
| 17 Nov, 22:18 UTC | 2% |
| 17 Nov, 19:55 UTC | 2% |
| 17 Nov, 19:28 UTC | 2% |
| 17 Nov, 18:20 UTC | 2% |
| 17 Nov, 15:52 UTC | 2% |
| 17 Nov, 15:06 UTC | 2% |
| 17 Nov, 13:56 UTC | 2% |
| 17 Nov, 13:07 UTC | 2% |
| 17 Nov, 12:52 UTC | 2% |
| 17 Nov, 11:44 UTC | 2% |
| 17 Nov, 11:26 UTC | 2% |
| 17 Nov, 09:53 UTC | 1% |
| 17 Nov, 08:47 UTC | 1% |
| 17 Nov, 07:13 UTC | 1% |
| 17 Nov, 05:47 UTC | 1% |
| 17 Nov, 04:22 UTC | 1% |
| 17 Nov, 00:53 UTC | 1% |
| 16 Nov, 21:48 UTC | 2% |
| 16 Nov, 18:22 UTC | 1% |
| 16 Nov, 17:43 UTC | 2% |
| 16 Nov, 17:01 UTC | 2% |
| 16 Nov, 16:54 UTC | 2% |
| 16 Nov, 16:25 UTC | 2% |
| 16 Nov, 15:44 UTC | 2% |
| 16 Nov, 13:27 UTC | 2% |
| 16 Nov, 12:38 UTC | 2% |
| 16 Nov, 12:05 UTC | 2% |
| 16 Nov, 03:41 UTC | 2% |
| 16 Nov, 03:21 UTC | 2% |
| 15 Nov, 20:20 UTC | 1% |
| 15 Nov, 19:53 UTC | 2% |
| 15 Nov, 17:05 UTC | 2% |
| 15 Nov, 16:04 UTC | 2% |
| 15 Nov, 15:29 UTC | 2% |
| 15 Nov, 14:48 UTC | 1% |
| 15 Nov, 14:17 UTC | 1% |
| 15 Nov, 13:28 UTC | 2% |
| 15 Nov, 12:54 UTC | 1% |
| 15 Nov, 10:27 UTC | 2% |
| 15 Nov, 03:04 UTC | 1% |
| 15 Nov, 02:18 UTC | 2% |
| 14 Nov, 20:40 UTC | 2% |
| 14 Nov, 19:11 UTC | 2% |
| 14 Nov, 18:47 UTC | 1% |
| 14 Nov, 16:54 UTC | 2% |
| 14 Nov, 15:53 UTC | 1% |
| 14 Nov, 12:39 UTC | 1% |
| 14 Nov, 11:50 UTC | 1% |
Replay the odds, article by article
Step through each article in the order it published and watch Will OpenAI have the best AI model at the end of November 2026? respond.
The story so far
8 publishers have covered this in 10 articles since 17 Nov 2026. theinformation.com published first, on 17 November at 05:03 UTC; Bloomberg Technology followed 12h 42m later. The most recent article, from Latest from Tom's Hardware in News, appeared on 20 November at 10:30 UTC.
Outlets reporting it: Bloomberg Technology (major outlet), theinformation.com, Quartz, decrypt.co, marktechpost.com, OpenAI, cnet.com, Latest from Tom's Hardware in News.
3 prediction markets price this story:
- Will OpenAI have the best AI model at the end of November 2026? — the market puts it at 1%, with $4k traded in the last 24 hours, closing 1 Dec 2026. The market resolves to Yes if OpenAI owns the model with the highest rank on the arena.ai Text Arena leaderboard at noon ET on November 30, 2026. The resolution relies strictly on the official leaderboard data from that specific source and time, ignoring any models marked as AutoEval. Readers often mistake the deadline for the end of the month or assume other ranking systems determine the winner.
- Will the highest score achieved by an OpenAI model on Humanity’s Last Exam in 2026 be 55% or higher? — the market puts it at 31%, closing 1 Jan 2027. The Yes outcome occurs if an OpenAI model reaches at least 55% accuracy on Humanity’s Last Exam by December 31, 2026, at 11:59 PM ET. The No outcome applies if no such score appears by that deadline or if the official leaderboard remains unavailable. Readers often mistake the linked news story about GPT-6.2 for the resolution data, but only the specific HLE Accuracy metric on the official Safe AGI leaderboard counts.
- Will OpenAI have the best AI model on LiveBench (Coding) at the end of December 2026? — the market puts it at 34%, with $80 traded in the last 24 hours, closing 1 Jan 2027. The outcome depends on which company owns the model with the highest Coding score on LiveBench.ai at 12:00 PM ET on December 31, 2026. If scores tie, the model with the lower cost per successful task wins; if costs also tie, alphabetical order of company names decides the winner.
Odds are the market's probability that the outcome happens, updated every few minutes. They are information about what traders expect, not a forecast we make and not investment advice.
How it unfolded
Every source in the order it published, with the odds on Will OpenAI have the best AI model at the end of November 2026? at that moment. A dash means the source published before we started tracking prices.
- theinformation.comOpenAI’s Astra Is a Boon to Robotics, Says Nvidia Robotics Chief Deepu Talla1%
- Bloomberg TechnologyOpenAI Unveils New Astra AI Model After Scrapping Prior Release2%
- QuartzOpenAI is unveiling its most advanced AI model after scrapping a prior version2%
- decrypt.coOpenAI's GPT 6.2 Astra Challenges Anthropic on Price, But Not Every Benchmark2%
- marktechpost.comOpenAI Releases GPT-6.2 Astra and Sol With a 20% Cheaper Astra API2%
- OpenAIAdvancing chip design with OpenAI models2%
- decrypt.co'It Speaks in Tongues That Only 6.2 Astra Understands': Why Early Testers Hate OpenAI’s New Model2%
- theinformation.comOpenAI Says Its Models Accelerated Chip Design For AMD, Jalapeño1%
- cnet.comOpenAI’s GPT-6.2 and Sol Are Getting Backlash From Early Testers1%▼1 points since the previous source
- Latest from Tom's Hardware in NewsOpenAI helped AMD improve its upcoming flagship GPU design by 14% PPA — GPT models also helped Lisa Su's outfit boost its new CPUs1%
- Current odds1%
Sources
Every article we clustered into this story. Headlines link to the publisher.
- OpenAI Unveils New Astra AI Model After Scrapping Prior Release Bloomberg Technology ·
- OpenAI helped AMD improve its upcoming flagship GPU design by 14% PPA — GPT models also helped Lisa Su's outfit boost its new CPUs Latest from Tom's Hardware in News
- OpenAI’s GPT-6.2 and Sol Are Getting Backlash From Early Testers cnet.com
- OpenAI Says Its Models Accelerated Chip Design For AMD, Jalapeño theinformation.com
- 'It Speaks in Tongues That Only 6.2 Astra Understands': Why Early Testers Hate OpenAI’s New Model decrypt.co
- OpenAI Releases GPT-6.2 Astra and Sol With a 20% Cheaper Astra API marktechpost.com
Show 4 more sources
- OpenAI's GPT 6.2 Astra Challenges Anthropic on Price, But Not Every Benchmark decrypt.co
- OpenAI is unveiling its most advanced AI model after scrapping a prior version Quartz
- OpenAI’s Astra Is a Boon to Robotics, Says Nvidia Robotics Chief Deepu Talla theinformation.com
- Advancing chip design with OpenAI models OpenAI
Related prediction markets
Linked by software because they are about the same subject. Most ask a different question from the one this story reports, so read each market's own question. How often the link is right
- Polymarket Will OpenAI have the best AI model at the end of November 2026? 1%
Does the market have this right?
Wrong market?
- Polymarket Will the highest score achieved by an OpenAI model on Humanity’s Last Exam in 2026 be 55% or higher? 31% −13
Does the market have this right?
Wrong market?
- Polymarket Will OpenAI have the best AI model on LiveBench (Coding) at the end of December 2026? 34% +3
Does the market have this right?
Wrong market?
Have your say
Your call. How does this end?
Free, no money involved. Calls are scored against the market's price when you make them, so backing an underdog that wins scores most. Leaderboard · Your calls