[KongchangAI]
Unverified50% confidenceBenchmarkExact time

Lowering Opus's effort setting meant the same three tasks used only 8% and 9% of the quota (roughly halved), with output volume nearly halved, list-price cost of $1.11 in both runs, and about 20 response steps instead of around 40, with all six passing.

1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/10/2026
First Seen
Valid until: 2/8/2027

Sources

Related Entities

Related Claims

Cite This Claim

Stable URI
https://kongchang.com/claim/1155637
API
curl https://kongchang.com/api/v1/knowledge/claims/1155637
MCP
get_claim(id=1155637)