Unverified50% confidenceBenchmarkExact time
On the AIME24 and AIME25 benchmarks, the reward budget mechanism achieved improvements of +10.00 each.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/5/2026
First Seen
Valid until: 2/3/2027
Sources
Related Entities
Related Claims
UnverifiedA 20% price reduction means that the amount of text that can be processed with the same budget increases by 25%.69% similarUnverifiedAgent inference is 40-100 times cheaper at list price, but after factoring in escalation, remediation, and goodwill loss, the real cost gap narrows to 20-30 times.65% similarUnverifiedWith the corruption process fixed, the choice of sampler produced a 25.0 percentage point performance gap at a budget of 180 calls.64% similarUnverifiedIn this cost breakdown, model replies account for 25.5% of total cost, ranking second.63% similarUnverifiedOn August 24, Cursor increased included usage without disclosing the magnitude of the increase and changed Auto billing from a flat rate to per-model billing; Z.ai changed its coding plan from prompt counts to token quotas; Atlassian re-measured automations by step count.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1137751API
curl https://kongchang.com/api/v1/knowledge/claims/1137751MCP
get_claim(id=1137751)