Unverified50% confidenceFactExact time
REKI 2.4B has 2.4B total parameters, consisting of a 2.2B-parameter vision-language backbone and a 200M-parameter Flow Matching action expert.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedREKI uses Flow Matching to generate actions and uses 10-step Euler integration at inference time.63% similarUnverifiedOn three out-of-distribution benchmarks, only models with roughly twice REKI's parameter count were able to surpass REKI.62% similarUnverifiedRebuff is a self-hardening prompt injection detection framework based on dual validation using heuristic rules and an LLM.60% similarUnverifiedREKI pretraining preserves each dataset's native control convention, and each robot embodiment enters the action expert through a dedicated state MLP plus AdaRMSNorm.60% similarUnverifiedThe three core elements of a custom ReAct agent are an LLM model (as the reasoning engine), Tools, and an observation loop.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1070494API
curl https://kongchang.com/api/v1/knowledge/claims/1070494MCP
get_claim(id=1070494)