Unverified50% confidenceFactExact time
GLOBE's probing branch executes in parallel with the main process, so the probing cost can be masked by the generation process, which is why the E2E reduction (51.0%) is slightly higher than the token reduction (49.7%).
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGLOBE achieves a 49.7% reduction in output tokens, a 51.0% reduction in end-to-end actual elapsed time (E2E), and an average accuracy of 80.58% across the five tasks, comparable to full CoT.73% similarUnverifiedGoogle 内部统计数据表明，E2E 测试与单元测试的数量比例建议维持在 1:10 乃至更低69% similarUnverifiedThe cost study results confirm that the optimized branch requires significantly fewer tokens to complete the same task.64% similarUnverifiedA developer ran a one-month tracking experiment on Claude Code's token consumption and found that the weighted cost ceiling of almost all optimization methods was below 2%.63% similarUnverifiedWhen completing the same tasks, SWE-2 requires 58% fewer interaction turns than its predecessor SWE-1.7, costs 81% less, and achieves a higher score.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1074056API
curl https://kongchang.com/api/v1/knowledge/claims/1074056MCP
get_claim(id=1074056)