Unverified50% confidenceBenchmarkExact time
Using only about 6% of Qwen2.5-VL-7B's parameter count, GUI-JEPA reached 86% to 92% of the latter's in-distribution accuracy on AndroidControl.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/28/2026
First Seen
Valid until: 1/26/2027
Sources
Related Entities
Related Claims
UnverifiedDirect use of Kev achieves 64.00% accuracy, and Qwen3.5-4B outputting only the final answer achieves 68.80% accuracy.66% similarUnverifiedOn the held-out compound recipe test set, the LoRA fine-tuned model caught 66%, while the base Qwen3.5-4B model caught 78%.64% similarUnverifiedOn VL-RewardBench, JEV-27B-VL achieves an overall accuracy of 78.3%, 83.2% on hallucination items, 78.2% on reasoning items, and 58.0% on general questions.64% similarUnverifiedLuna scored 66.6% in the coding test, with costs 93% and 96% lower than Opus 5 and Fable 5, respectively.64% similarUnverifiedOn "continuation of previous operation" type utterances, Qwen3.6-35B-A3B achieved an accuracy of 86.3%, while the 4B and 8B models achieved only 70.4%.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1106119API
curl https://kongchang.com/api/v1/knowledge/claims/1106119MCP
get_claim(id=1106119)