Agent-0 training reaches 10²⁷ FLOP
OpenBrain’s latest public model—Agent-0—was trained with 10²⁷ FLOP.
Late 2025: The World’s Most Expensive AI · Archived PDF, p. 4
The May 2025 completion milestone remains unresolved. Epoch’s later Astra estimate is 9.4 × 10²⁶ FLOP with an interval spanning 10²⁷, not proof of a crossing or its completion date. Behind 60% remains tentative.
At a glance
- Assessment: Behind
- Confidence in assessment: 60%
- Outcome: unresolved
- Timing: Original window passed; outcome unresolved
- Evidence: proxy
- Predicted timing: Training completion: May 2025; public availability: late 2025
- Primary source: Late 2025: The World’s Most Expensive AI, archived PDF p. 4
What AI 2027 Predicted
Tracker summary: The compute supplement models Agent-0 training from October 2024 through May 2025 at 10²⁷ FLOP. Public availability later in 2025 is a separate milestone. The May date therefore describes completion of a run already underway when the scenario was published, rather than a new project to be started and finished the following month. AI 2027 Compute Forecast, section 5
Scenario source locator: Late 2025: The World’s Most Expensive AI: archived PDF, p. 4.
How We Track This
Compare estimated compute used in individual frontier training runs with the modeled scale and completion window. A release date is an imperfect proxy for training completion. Public estimates cannot rule out undisclosed runs.
Current Evidence
Later estimate, not a proven crossing: Epoch’s model dataset estimates GPT-6 Astra at 9.4 × 10²⁶ FLOP, with an approximate interval of 5 × 10²⁶ to 2 × 10²⁷ FLOP in its training-compute notes. The central estimate is near, but below, 10²⁷ and the interval spans the threshold. This is original research reconstruction, not company training accounting. Its September 3, 2026 publication field is not a training-completion date. Neither an on-time May 2025 result nor a later crossing is established; Behind (60%) remains a tentative trajectory assessment.
For the historical 2025 comparison, Epoch’s model dataset estimates Grok 4 at 5 × 10²⁶ FLOP, GPT-4.5 at 3.8 × 10²⁶ and Grok 3 at 3.5 × 10²⁶. These characterized 2025 runs were approximately two to three times below the target. They are much closer than a comparison using GPT-5 alone would suggest. These values are research reconstructions, not company-confirmed training accounts. Epoch model dataset, retrieved November 7, 2026. Epoch revises historical estimates; the values above describe this retrieval, not a fixed company disclosure.
Epoch’s October 2025 GPT-5 analysis estimated 5 × 10²⁵ FLOP. That dated estimate remains part of the historical record; it does not establish the maximum compute used by any lab. Epoch’s GPT-5 analysis
Counterevidence & Limitations
The characterized 2025 estimates support a tentative historical scale shortfall; the later Astra interval spans the target. Model identities, training boundaries and unreported experiments remain uncertain. Comparisons with models released after May cannot establish what had completed by May. Changes in training efficiency also mean a compute shortfall need not imply a proportionate capability shortfall.
Behind expresses a tentative assessment of the modeled training trajectory. It is not a confirmed failure established one month after publication.
What Would Change Our Assessment
- Strengthen: Comparable records place a completed run at or above 10²⁷ FLOP, with its completion date documented.
- Confirm a shortfall: More complete lab records establish that the relevant frontier remained materially below the target within the window.
- Reassess timing: A later qualifying run would distinguish a delayed milestone from an unobserved one.
Update History
| Date | Update |
|---|---|
| 2026-11-09 | Assessment reviewed. Status: Behind; confidence: 60%. Added Epoch’s later Astra estimate of 9.4 × 10²⁶ FLOP with an approximate 5 × 10²⁶ to 2 × 10²⁷ interval. Neither a threshold crossing nor a completion date is established. May 2025 completion and later publication remain distinct. |
| 2026-11-07 | Assessment reviewed. Status: Behind; confidence: 60%. Epoch’s model dataset, retrieved November 7, places the leading characterized 2025 runs at roughly 3.5 to 5 × 10²⁶ FLOP, below 10²⁷. This supports a tentative scale shortfall, not a proven May failure: estimates can be revised, release dates are not completion dates, and undisclosed runs remain possible. Clarified the distinction between May training completion and later public release; dated the revisable Epoch estimates. The September 7 assessment correction remains Behind (55%) to Behind (60%). |
| 2026-09-11 | Clarification of the September 7 comparison: leading characterized 2025 runs were approximately 2-3× below the modeled target. That broader comparison set confidence at 0.60, following the earlier 0.55 assessment recorded below. The current Behind rating and 0.60 confidence are unchanged. |
| 2026-09-07 | Expanded the comparison beyond GPT-5: leading characterized 2025 runs were approximately 2-3× below the target. May training completion and later public availability remain separate; the exact milestone is unresolved. 2026-11-07 · Correction: status Behind (55%) → Behind (60%). Leading characterized 2025 runs were about 2 to 3 times below the target. |
| 2026-09-07 | Distinguished May training completion from late-2025 public availability and clarified the source of both dates. Behind remains a tentative trajectory assessment; the exact milestone is unresolved. Assessment confidence reduced from 0.75 to 0.55. |
| 2026-09-06 | Assessment revised from emerging (0.45) to behind (0.75). No qualifying run was established by the original May 2025 deadline; estimates of disclosed runs do not exclude undisclosed training. |
| 2026-03-13 | GPT-5 estimated at ~3×10²⁵ FLOP pretraining, well below 10²⁷ target. Significant uncertainty remains around total compute including post-training and inference-time scaling. |