Unverified50% confidenceFactExact time
21.2% is a relative reduction rather than an absolute one, meaning the new version eliminates approximately one-fifth of the failures relative to the earlier version's failure rate baseline.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/25/2026
First Seen
Valid until: 12/24/2026
Sources
Related Entities
Related Claims
Unverified在SWE-bench Full版本上，多数模型的通过率仍低于20%74% similarUnverifiedIn real-environment A/B testing, the post-trained checkpoint reduced the tool-call failure rate by 21.2% relatively compared to the earlier version.73% similarVerifiedIn long-chain tasks, if each step has an error rate of approximately 5%, the overall success rate may drop to below 36% after 20 steps.73% similarUnverifiedThe current best-performing model on MNIST has an error rate below 0.2%.71% similarUnverifiedIn 446 pairwise comparisons, CTD performed comparably in 21.1% of cases and was statistically significantly inferior to its competitors in only 6.1% of cases.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/948596API
curl https://kongchang.com/api/v1/knowledge/claims/948596MCP
get_claim(id=948596)