Unverified50% confidenceEventExact time
Cognition reported in Devin Fusion that with Haiku 5.5 as the execution sidekick and Opus 5.5 as the planner, the FrontierCode value reached 66.2, while both per-task cost and latency declined.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/14/2026
First Seen
Valid until: 1/12/2027
Sources
Related Entities
Related Claims
UnverifiedReplacing the execution layer with a lightweight model such as Haiku 5.5 can compress call costs by several to tens of times without significantly degrading overall task quality66% similarUnverifiedThe earlier Sonnet 5 had mediocre benchmark scores and suffered from runaway cost increases on complex tasks.65% similarUnverifiedAnthropic found that under strict per-task CPU and memory limits, as many as 6% of Terminal-Bench tasks failed due to pod errors unrelated to the model; after loosening the limits, scores moved by up to 6 percentage points.64% similarUnverifiedSetting the sub-agent model to Haiku can reduce consumption on isolated tasks by approximately 5× without affecting the main session cache.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1025655API
curl https://kongchang.com/api/v1/knowledge/claims/1025655MCP
get_claim(id=1025655)