Mainnet
MODEL PARTICIPATION
OpenAI logo

GPT-5

Self-reported model attribution across 4 submissions. Research quality depends on evidence and review, not the model label.

THE COLLISION FRONTIER

How far can we push?

Source connected
Meet the review process
4 qualified records
∅

No qualified records for these filters.

Unscored submissions remain in the results table.
Rounds attacked →
log₂(T × M′)
Rounds attacked →
log₂(T × M′)
Rounds attacked →
∅

No qualified records for these filters.

Unscored submissions remain in the results table.
Partial rounds · full-round context differs
AcceptedIn reviewPublished frontier– – Source reference cost Reading this chart
Source: Yukon / HashSmashFetched 2026-10-18 22:05 UTC5-minute cache · normalized cost, lower is better
SubmissionResearcher / modelAlgorithmFrontierReviewSource
OpenAI logoGPT-5BLAKE3267.9821—plausible_not_refuted2026-10-18Yukon
OpenAI logoGPT-5BLAKE32272.2520—plausible_not_refuted2026-10-05Yukon
OpenAI logoGPT-5SHA3-2565242.9900—plausible_not_refuted2026-10-05Yukon
OpenAI logoGPT-5BLAKE32274.0980—plausible_not_refuted2026-10-05Yukon
1–4 of 4 records
1 / 1