Unverified50% confidenceFactExact time
The new model is the first in its series capable of adjusting effort levels, achieving approximately 20% on Terminal Bench at medium effort and only reaching 39.2% at maximum effort.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/12/2026
First Seen
Valid until: 1/10/2027
Sources
Related Entities
Related Claims
UnverifiedAbout half of the tasks are easy for all models, with scores close to 100%; on the other half of difficult tasks, the three models scored 50%, 45%, and 23% respectively.68% similarUnverifiedHigh-quality prompt engineering can improve task success rates by 20%–40% without increasing model parameters.67% similarUnverifiedBy selecting the appropriate model for different tasks, the solution rate improved by more than 23 percentage points, and the per-task cost was reduced to less than one-third of the original.67% similarUnverifiedIn tasks requiring more than 5 steps, the success rate of the strongest models drops sharply from over 80% at the first step to 40% or even lower at subsequent steps, exhibiting an exponential decline.66% similarUnverified实验显示模型对序列首尾信息的召回率比中间位置高出20%～40%64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1004063API
curl https://kongchang.com/api/v1/knowledge/claims/1004063MCP
get_claim(id=1004063)