Unverified50% confidenceBenchmarkExact time

DeReAct improves the Pass@1 of Claude Sonnet 4.5 by 4.2 to 5.2 percentage points.

1

Sources

50%

Confidence

Long-term

Relevance

10/5/2026

First Seen

Sources

DeReAct: Decoupling Reasoning and Acting for More Reliable AI Agents

Related Entities

DeReActconcept Claude Sonnet 4.5model

Related Claims

UnverifiedModels represented by Claude 3.5 Sonnet have reached engineering-practical levels in code generation pass@1 rates.70% similar UnverifiedDeReAct improves the Pass@1 of Qwen3-Coder-480B by 6.5 to 7.0 percentage points.69% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/975646

API

curl https://kongchang.com/api/v1/knowledge/claims/975646

MCP

get_claim(id=975646)