Unverified50% confidenceBenchmarkExact time
The SlopCodeBench October test report shows that after adjusting reasoning effort from medium to xhigh, cost rose to 2 to 4 times, while the number of checkpoints passed did not improve reliably.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/21/2026
First Seen
Valid until: 11/4/2026
Sources
Related Entities
Related Claims
UnverifiedIn the Skatebench test, from the x-high tier to the max tier, reasoning tokens per question spiked from approximately 330 to nearly 5,000 (more than 10x), while scores improved by only 1%.67% similarUnverifiedOn checkpoint pass rate, xhigh showed no statistically reliable improvement; on some tasks the two were on par, and on individual tasks xhigh was even slightly lower.64% similarUnverifiedThe marginal returns from moving from the High to Max inference tier are clearly diminishing: the DeepSWE score increases by only approximately 4 percentage points, while the cost increases by approximately 1.5x.64% similarUnverified研究数据显示，在复杂代码生成任务中，将thinking budget从1024 tokens提升到8192 tokens可以带来15-30%的准确率提升，但推理成本增加4-8倍63% similarUnverifiedAutomated evaluation shows that DiscoveryRSI's hypothesis quality rose from 4.86 for the base model to 5.39 after two rounds of self-improvement, with gains concentrated in the falsifiability and data-parseability dimensions.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1081131API
curl https://kongchang.com/api/v1/knowledge/claims/1081131MCP
get_claim(id=1081131)