Unverified50% confidenceBenchmarkExact time
In the evaluation with seven held-out queries, the trained IRM achieved 4/7 correct on top-1 results, versus 3/7 for the identity-transform baseline.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/21/2026
First Seen
Valid until: 11/4/2026
Sources
Related Entities
Related Claims
UnverifiedOn the identity-related subset, IRM achieved 3/5 correct, versus 1/5 for the identity-transform baseline.76% similarUnverifiedIn Artificial Analysis's AA-Omniscience benchmark (which penalizes incorrect answers), the top score rose from 4.8 at launch to 46 currently.66% similarUnverifiedThe final parameter count of the model after training was 13,693,824.66% similarUnverifiedOnce GOOSE is included in the training set, all augmentation branches perform within ±1 point of the control group on real weather, while augmentation still adds 3.5 to 5.2 points on synthetic held-out degradations.65% similarUnverifiedBilibili UP host 'Evolving Acheng' evaluated GPT-6 Sol using a benchmark of 9 major questions, yielding a weighted total score of 95 points.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1081277API
curl https://kongchang.com/api/v1/knowledge/claims/1081277MCP
get_claim(id=1081277)