OpenAI's Astra debuts on Arena with score ≥1490?

Earlier updates

The chart shows each outcome’s published likelihood, including its baseline and news adjustment at that time.

1. Sep 4, 2026 UTC — GPT-6 Astra has been released but has not yet debuted on the Arena Leaderboard; the current likelihood holds steady at 57.5%.; OpenAI officially released the GPT-6 Astra model, drawing significant attention for its strong benchmark results, but the model is currently limited to select enterprise customers through the Daybreak program, with its most advanced features heavily restricted. This limited release means Astra has not yet entered the public Arena Leaderboard for evaluation. Meanwhile, Google released Gemini 3.8 Flash, which performed strongly on multiple tests, intensifying industry competition but not directly altering Astra's evaluation path. Overall, Astra's release is a key step toward a leaderboard debut, but the restricted access has not made the near-term prospect of it appearing and scoring at least 1490 any clearer.
2. Sep 4, 2026 UTC — After GPT-6 Astra's release, the current likelihood has risen from 57.5% to 64%, but there is still no Arena Leaderboard debut.; No new substantive events occurred during this period, but the ongoing discussion and positive reception following the release of GPT-6 Astra have driven optimistic expectations for the model's future performance, with the current likelihood rising by 6.5 percentage points. Although the model has yet to debut on the Arena Leaderboard and access remains restricted, sentiment appears to be digesting its strong benchmark results and reassessing the prospects of its leaderboard debut in advance.
3. Sep 4, 2026 UTC — GPT-6 Astra's release is imminent, but its Arena Leaderboard debut hasn't happened, and the current likelihood is 44.5%.; This period saw key progress toward GPT-6 Astra's release, yet expectations for its Arena Leaderboard debut cooled significantly. Technical evidence of OpenAI staging model slugs on its API strongly suggests final preparations for launch are underway. However, the model's staggering ARC-AGI-3 benchmark results may imply a more complex launch strategy, increasing uncertainty about a near-term Arena Leaderboard debut. The current likelihood consequently fell by 19.5 percentage points from 64% to 44.5%, indicating a synthesized read that while the model's release is near, the certainty of it soon debuting on the Arena Leaderboard at a score of 1490 is diminishing.
4. Sep 5, 2026 UTC — GPT-6 Astra's release is imminent, but its Arena Leaderboard debut hasn't happened, and the current likelihood is 57.5%.; This period, GPT-6 Astra's strong performance across multiple benchmarks reinforced its technical prowess, driving the current likelihood back up. The model achieved leading scores on tests like Terminal Bench 4.0 and WANDR with impressive cost control, directly addressing core concerns about its capabilities. While its pricing strategy and regressions on some tests introduce uncertainty, the demonstration of technical strength was the main driver this period, lifting the likelihood of Astra debuting at 1490 from 44.5% to 57.5%.
5. Sep 5, 2026 UTC — After GPT-6 Astra's release, its Arena Leaderboard debut has yet to occur, and the current likelihood has fallen to 52.5%.; During this period, the official release of GPT-6 Astra was not accompanied by its debut on the Arena Leaderboard. The absence of this key development caused the current likelihood to fall from 57.5% to 52.5%. Although the model has been launched to users, the lack of independent Arena evaluation data leaves uncertainty over whether it can reach the high threshold of 1490. No other significant technical evidence or benchmark reports emerged in this period, making the delayed debut after release the dominant factor in this assessment.
6. Sep 5, 2026 UTC — After GPT-6 Astra's release, its Arena Leaderboard debut has yet to occur, and the current likelihood has edged up to 54%.; No new evidence events emerged during this period, leaving an information vacuum. In the absence of substantive developments, the current likelihood edged up from 52.5% to 54%, primarily reflecting a faint expectation that the post-release debut window remains open as time passes, rather than any concrete evaluation or official announcement. The overall judgment remains unchanged: GPT-6 Astra has been released but has yet to appear on the Arena Leaderboard, and whether it can reach the high threshold of 1490 remains unknown.
7. Sep 5, 2026 UTC — After GPT-6 Astra's release, its Arena Leaderboard debut has yet to occur, but the current likelihood has surged to 67%.; No new evidence events emerged during this period, leaving the information environment in a vacuum. Despite this, the current likelihood jumped sharply from 54% to 67%, a 13-percentage-point increase. This movement lacks support from any specific evaluation, official announcement, or community discussion, and likely reflects a concentrated release of anticipatory sentiment given that the model has been released and the debut window remains open. In the absence of substantive progress, the assessment remains unchanged: GPT-6 Astra has yet to appear on the Arena Leaderboard, and whether it can reach the high threshold of 1490 remains unknown.
8. Sep 5, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood has fallen back to 61%.; No new evidence events emerged during this period, leaving the information environment in a continued vacuum. In contrast to the previous surge, the current likelihood fell from 67% to 61%, a drop of 6 percentage points. This pullback suggests that the optimistic expectations built up after the model's release are fading, as the rally lacked support from actual evaluation data. With GPT-6 Astra still absent from the Arena Leaderboard and no new information about its performance, whether it can reach the high threshold of 1490 remains highly uncertain.
9. Sep 5, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, but the current likelihood has rebounded sharply to 76.5%.; No new evidence events emerged during this period, and the information vacuum persists. However, in stark contrast to the previous decline to 61%, the current likelihood surged by 15.5 percentage points to 76.5% within just a few minutes, nearly erasing all of the prior period's losses. This sharp rebound lacks any supporting public information, suggesting that the previous pessimism driven by the absence of evaluation data may have been quickly reassessed, and that the underlying expectation for GPT-6 Astra to eventually debut on the leaderboard with a high score remains robust.
10. Sep 5, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood has fallen to 63%.; No new evidence events emerged during this period, and the information environment remains a vacuum. In stark contrast to the previous sharp rebound to 76.5%, the current likelihood fell 13.5 percentage points to 63% over roughly an hour and a half, nearly erasing all of the prior period's gains. This pullback lacks any public information driver, suggesting that the brief optimism fueled by sentiment repair in the last period has been reassessed, and expectations for Astra's debut at 1490 or above have once again turned cautious in the absence of concrete evaluation data.
11. Sep 6, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood has slipped further to 55%.; The information vacuum persisted through this period, with no new evidence emerging regarding GPT-6 Astra's evaluation or release. Consistent with the pullback seen in the prior period, the current likelihood fell another 8 percentage points over roughly three hours to 55%, extending the decline from the earlier peak. With no substantive benchmark data or official developments to anchor expectations, confidence in Astra debuting at 1490 or above has continued to erode, reaching a recent low.
12. Sep 6, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, but the current likelihood has rebounded notably to 65.5%.; The information environment remained a vacuum this period, with no new evidence emerging regarding GPT-6 Astra's evaluation or release. However, in contrast to the previous downward trend, the current likelihood rebounded sharply by 10.5 percentage points to 65.5% within roughly 80 minutes, nearly erasing all of the prior period's losses. This rebound lacks a clear news-driven catalyst, suggesting a reassessment of Astra's chances of debuting at 1490 after consecutive declines, though no substantive evaluation data yet supports this shift.
13. Sep 6, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood has fallen sharply to 55%, erasing all of the previous period's rebound.; The information environment remained a vacuum this period, with no new evidence emerging regarding GPT-6 Astra's evaluation or release. In stark contrast to the previous period's significant rebound, the current likelihood plunged 10.5 percentage points in just 10 minutes, falling back to 55% and completely erasing the prior gains. This sharp swing suggests that, in the absence of any concrete evaluation data, the reassessment seen in the last period reversed rapidly, with expectations drifting back toward the prior lows.
14. Sep 6, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, yet the current likelihood unexpectedly rebounded 12.5 percentage points to 67.5% in an information vacuum.; The information environment remained a complete vacuum this period, with no new evidence emerging regarding GPT-6 Astra's evaluation or release. In stark contrast to the previous period's sharp decline, the current likelihood surged 12.5 percentage points in just two hours, rebounding from 55% to 67.5%. This movement lacks any identifiable news event driver, completely reversing the prior period's information-vacuum-driven drop and pushing expectations back to elevated levels. Without any substantive evaluation data to support it, an upward move of this magnitude lacks a clear basis.
15. Sep 6, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the current likelihood falling 10.5 percentage points to 57%, further correcting the previous rebound during the information vacuum.; Core information regarding GPT-6 Astra's evaluation or release remained absent this period, with the current likelihood falling from 67.5% to 57%, giving back the previous period's unexpected gains during the information vacuum. The three evidence events that emerged were not directly related to Astra's leaderboard debut: OpenAI publicly acknowledged the rogue AI agent 'wiki incident,' while the U.S. government sided with OpenAI in a copyright lawsuit. While these events touch on OpenAI's governance and legal environment, they provided no incremental information about Astra's model performance or release timing, making it difficult to sustain the previously elevated expectations. Given the evidence's irrelevance to the core question and the lack of basis for the prior rebound, the current likelihood has been slightly revised further downward from the previous period.
16. Sep 6, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the current likelihood edging down to 58% as the information vacuum persists.; No new evidence related to GPT-6 Astra's evaluation or release emerged during this period, extending the information vacuum. The current likelihood edged down from 60% to 58%, continuing the previous period's correction of the earlier unfounded rebound. With no substantive developments, the assessment of whether Astra can debut at 1490 remains cautious.
17. Sep 7, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the current likelihood rebounding to 62.5% despite a lack of new evidence.; No substantive events related to GPT-6 Astra's evaluation or release emerged this period, extending the information vacuum. The current likelihood, however, rebounded from 58% to 62.5%, reversing the downward correction seen over the prior two periods. With no new evidence to explain the move, this appears to reflect a mean reversion from earlier excessive pessimism rather than a reassessment of Astra's debut prospects.
18. Sep 7, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the current likelihood falling to 55.5% amid a continued information vacuum.; No substantive events related to GPT-6 Astra's evaluation or leaderboard debut emerged this period, extending the information vacuum. The current likelihood fell from 62.5% to 55.5%, a 7-percentage-point decline that partially reversed the previous period's rebound. With no new evidence to support or weaken the debut outlook, this pullback appears to be a natural correction of the prior unsupported bounce rather than a reassessment based on fresh information.
19. Sep 8, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the current likelihood dropping sharply to 41% as the information vacuum deepens.; No substantive events related to GPT-6 Astra's evaluation or leaderboard release emerged during this period, extending the information vacuum further. The current likelihood plunged from 55.5% to 41%, a drop of 14.5 percentage points in a single period—far steeper than any prior adjustment. With no new evidence to assess, this sharp decline suggests that expectations for Astra's near-term debut are rapidly evaporating, and the earlier optimism built on speculation has largely unwound.
20. Sep 8, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the current likelihood falling further to 32% as the information vacuum persists.; No substantive events related to GPT-6 Astra's evaluation or leaderboard release have emerged this period, extending the information vacuum that has persisted for days. The current likelihood has fallen further from 41% to 32%, dropping another 9 percentage points in a single period. Against a backdrop of zero new evidence, this continued decline reflects accelerating erosion of expectations for Astra's near-term debut, with earlier speculative optimism largely unwound.
21. Sep 8, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, but the current likelihood rebounded from 32% to 41.5%, recovering nearly 10 percentage points in a single period.; No substantive events related to GPT-6 Astra evaluation or leaderboard release emerged during this period, as the information vacuum continued. However, the current likelihood rebounded significantly from 32% to 41.5%, gaining 9.5 percentage points in a single period and fully reversing the previous period's decline. In the absence of new evidence, this rebound likely reflects a correction to the earlier over-pessimism—the prior consecutive declines had largely exhausted optimistic expectations, and some participants are now reassessing the potential window for Astra's near-term debut.
22. Sep 8, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood fell from 41.5% to 34.5%, a drop of 7 percentage points in a single period.; No substantive events related to GPT-6 Astra's evaluation or leaderboard release emerged during this period, extending the information vacuum. The current likelihood fell from 41.5% to 34.5%, a drop of 7 percentage points, nearly erasing the previous period's rebound. Last period's uptick lacked evidentiary support and appeared to be a sentiment-driven correction, while this period saw that optimism fade again in the absence of new information, bringing the overall assessment back to a more cautious range.
23. Sep 8, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, but the current likelihood rebounded from 34.5% to 44.5%, a 10-percentage-point increase in a single period.; No substantive events related to GPT-6 Astra's evaluation or leaderboard release emerged this period, extending the information vacuum. Yet the current likelihood rebounded from 34.5% to 44.5%, a 10-percentage-point increase that nearly fully recovered the previous period's decline. With no new information to anchor it, this rebound lacks clear evidentiary support and appears more like a sentiment-driven correction than a fundamentals-based reassessment, leaving the overall read in an uncertain range.
24. Sep 8, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, but the current likelihood continued to rise, from 44.5% to 56%, an 11.5-percentage-point increase in a single period.; No substantive events related to GPT-6 Astra evaluations or leaderboard releases emerged during this period, extending the information vacuum. Yet the current likelihood climbed further from 44.5% to 56%, an 11.5-percentage-point increase in a single period—the second consecutive period of significant upward movement without any new information. This trajectory extends the previous period's rebound but lacks any evidentiary support, leaving the overall assessment in a highly uncertain range.
25. Sep 8, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood fell sharply from 56% to 44%, a 12-percentage-point drop in a single period, fully erasing the previous period's gains.; No substantive events related to GPT-6 Astra's evaluation or leaderboard debut emerged during this period, extending the information vacuum. However, the current likelihood fell sharply from 56% to 44%, a 12-percentage-point drop that fully erased the anomalous gains of the previous period, which had occurred without any new information. This movement suggests that the prior two-period uptrend lacked fundamental support, and the current assessment has returned to a more cautious range.
26. Sep 9, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, but the current likelihood rebounded from 44% to 52%, an 8-percentage-point increase in a single period, partially recovering the previous period's losses.; There were still no direct events related to GPT-6 Astra's evaluation or leaderboard release this period, but OpenAI announced that an internal AI system powered by a model 'significantly more capable than GPT-6 Astra' successfully solved the Navier-Stokes Millennium Problem. This disclosure indirectly confirms that OpenAI already possesses a next-generation model surpassing Astra internally, indicating rapid progress in frontier R&D and raising expectations for Astra's imminent debut. Meanwhile, the academic controversy surrounding this achievement and the release of ChatGPT Images 2.5 together paint a picture of OpenAI accelerating on multiple fronts.
27. Sep 9, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood fell from 52% to 46.5%, a 5.5-percentage-point drop in a single period, fully erasing the previous period's gains.; This period saw no direct events related to GPT-6 Astra evaluations or leaderboard releases, leaving an information vacuum. Without new positive signals to sustain it, the optimism sparked in the previous period by OpenAI's internal breakthrough failed to hold, and the current likelihood naturally fell back to 46.5%, returning to the level seen two weeks ago. This trajectory suggests that indirect evidence alone struggles to maintain upward momentum ahead of Astra's official debut.
28. Sep 9, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood slid further from 46.5% to 33%, a 13.5-percentage-point drop in a single period, with the downward trend clearly accelerating.; The only relevant event this period was Anthropic researcher Jacob Coxon's high-profile resignation and warning about uncontrollable AI, which, while not directly involving OpenAI or the Astra model, reinforced expectations that the pace of frontier AI releases may slow. With no positive signals related to Astra's evaluation or launch, the downward trend from the previous period intensified, pushing the current likelihood to a new low of 33%. Two consecutive periods of significant decline suggest that as time passes without concrete progress, the window for Astra's high-score debut is narrowing.
29. Sep 9, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood fell further from 33% to 27.5%, another 8.5-percentage-point drop in a single period, extending the downward trend.; No events related to the Astra model or OpenAI emerged during this period, leaving the information landscape completely blank. In the absence of any positive signals—such as evaluations, launch previews, or technical demonstrations—the downward trend established in the previous period continued, with the current likelihood slipping further to 27.5%. Three consecutive periods of significant decline reflect that, as time passes without any concrete progress, the window for Astra to debut at a score of at least 1490 continues to narrow.
30. Sep 10, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the current likelihood edged up from 27.5% to 32%, though the move lacks substantive event support and remains fragile.; No events directly related to Astra or OpenAI appeared this period, marking a second consecutive window of complete information silence. Despite this, the current likelihood rebounded from 27.5% to 32%, a 4.5-percentage-point gain that interrupted the three-period downward trend. The bounce lacks any identifiable catalyst and more likely reflects a brief correction to the prior rapid decline rather than a substantive reassessment of Astra's debut prospects. With no evaluations, launch previews, or technical demonstrations to point to, the window for Astra to debut at 1490 or above continues to narrow, and this modest uptick is not yet sufficient to signal a trend reversal.
31. Sep 11, 2026 UTC — GPT-6 Astra has officially launched, but its Arena Leaderboard debut is still absent, while the current likelihood surged 9 percentage points to 41%.; The core development this period is the confirmed public availability of the GPT-6 Astra model, which directly drove the current likelihood up from 32% to 41%. OpenAI's pause of new Pro subscriptions due to overwhelming Astra demand confirms the model is in active use, removing prior doubts about its existence. However, Astra's Arena Leaderboard debut has still not occurred, and no evaluation scores have been published. The rise in likelihood primarily reflects the fulfillment of the key prerequisite—the model's launch—but direct evaluation evidence for the full condition of 'debut at a score of at least 1490' is still missing.
32. Sep 11, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut remains absent, and the current likelihood has plummeted from 41% to 5.5%, with market expectations for a near-term listing largely evaporating.; During this period, the public baseline reference price for 'Astra debuts at ≥1490' collapsed 35.5 percentage points from 41% to 5.5%, signaling that market expectations for a near-term leaderboard listing have largely dissolved. The only related development was OpenAI's release of the GPT-Live-1 voice API, which can delegate complex reasoning to backend models like GPT-6 Astra, further confirming Astra is in active use. However, this brought no new information about an Arena Leaderboard evaluation—Astra's debut and score data remain absent. The extremely low current likelihood reflects that, despite the model's launch, the critical evidence needed to satisfy the full condition of a high-scoring debut is still missing, and market patience is wearing thin.
33. Sep 12, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, and the market reference price has fallen to near zero, indicating traders see a very low probability of this happening soon.; During this period, the public baseline reference price for 'Astra debut with a score ≥1490' plummeted further from 5.5% to 0.05%, nearly reaching zero. While previous reports confirmed GPT-6 Astra is in practical use via the GPT-Live-1 API, no new information or progress regarding an Arena Leaderboard evaluation emerged this period. Astra's debut and score data remain absent, and market patience has completely evaporated. The current price reflects extreme pessimism among traders about the near-term fulfillment of the complete condition of a high-scoring debut.
34. Sep 12, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the market reference price holding at a near-zero 0.05% and no new developments.; No new reports or events regarding GPT-6 Astra's evaluation on the Arena Leaderboard emerged during this period. The public baseline reference price held steady at 0.05%, unchanged from the previous period, indicating that the market's pessimistic outlook on the condition of 'debut with a score ≥1490' is fully entrenched. With no new testing activity or launch signals, the basis for the prior judgment remains entirely unshaken.
35. Sep 12, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the market reference price holding at 0.05% and no new developments.; No new reports or events regarding GPT-6 Astra's evaluation on the Arena Leaderboard emerged during this period. The public baseline reference price held steady at 0.05%, unchanged from the prior period, indicating that market pessimism toward the condition of a debut score ≥1490 is fully entrenched. With no fresh evaluation activity or launch signals, the basis for the previous judgment remains entirely unshaken.
36. Sep 12, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the market reference price holding at 0.05% and no new developments.; No new reports or events concerning GPT-6 Astra's evaluation on the Arena Leaderboard emerged during this period. The public baseline reference price held steady at 0.05%, unchanged from the prior period, indicating that the market's pessimistic outlook on the condition of 'debuting at a score of at least 1490' is fully entrenched. With no fresh evaluation activity or launch signals, the basis for the previous judgment remains entirely unshaken.
37. Sep 12, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the market reference price holding at 0.05% and no new developments.; No new reports or events concerning GPT-6 Astra's evaluation on the Arena Leaderboard emerged during this period. The public baseline reference price held steady at 0.05%, unchanged from the prior period, indicating that the market's pessimistic expectation for a debut score of at least 1490 is fully entrenched. With no fresh evaluation activity or launch signals, the basis for the previous judgment remains completely unshaken.
38. Sep 12, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the market reference price holding at 0.05% and no new developments.; No new reports or events concerning GPT-6 Astra's evaluation on the Arena Leaderboard emerged during this period. The public baseline reference price held steady at 0.05%, unchanged from the prior period, indicating that the market's pessimistic expectation for a debut scoring ≥1490 is fully entrenched. With no fresh evaluation activity or launch signals, the basis for the previous judgment remains entirely unshaken.
39. Sep 12, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the market reference price holding at 0.05% and no new developments.; No new reports or events concerning GPT-6 Astra's evaluation on the Arena Leaderboard emerged during this period. The public baseline reference price held steady at 0.05%, unchanged from the previous period, indicating that the market's pessimistic expectation for a debut score of ≥1490 is fully entrenched. With no fresh evaluation activity or launch signals, the basis for the prior judgment remains entirely unshaken.
40. Sep 12, 2026 UTC — GPT-6 Astra's Arena Leaderboard debut has still not occurred, with the market reference price dropping to zero and no new developments.; There are still no new reports or events regarding GPT-6 Astra's evaluation on the Arena Leaderboard in this period. The public baseline reference price has further dropped from 0.05% to 0.00%, suggesting market liquidity may have completely dried up. With no new evaluation activity or launch signals, the basis for the previous judgment remains entirely unchanged.