Expired50% confidenceBenchmarkExact time
In the "Legacy Bug Duel" test by Bilibili UP host「屎山论剑」, GPT-6 cleared all three difficulty levels — Gold, Diamond, and King — in a single round, scoring a perfect 8 points.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
9/21/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 10/5/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedIn the Diamond bug test, GPT-6 took 16 minutes but solved it correctly in one round for 3 points, Fable 5.1 fixed it successfully in the second round, and Gemini failed across all three rounds.72% similarUnverifiedFable 5.1 (Feibao) scored on both the Diamond and King bugs in that test, ultimately earning 5 points.72% similarUnverifiedAcross four rounds of testing, Astra won three and tied on dynamic graphics, leading the reviewer to judge GPT-6 as the overall winner.65% similarUnverifiedOn the Diamond subset, GPT-4o achieved a score of 6.3% and a pass rate of 7.2%.63% similarUnverifiedKimi K2.6 competes with closed-source models such as GPT 5.4, Claude Opus 4.6, and Gemini 3.1 Pro on top benchmarks including SWE Bench Pro, deep search question answering, and Humanity's Last Exam.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/939577API
curl https://kongchang.com/api/v1/knowledge/claims/939577MCP
get_claim(id=939577)