[Explosive AI Project J-Space Accused of Fraud, Community Retesting Results 'Completely Opposite' to Official Data] BlockBeats News, August 18 — According to MaxForAI, the AI project J-Space Cognition Suite, which went viral today on the X platform, has been questioned by the community, with claims that its advertised DeepSeek V4 test results are not reproducible. The project previously claimed that V4 Flash paired with J-Space could match GLM-5.3, while V4 Pro could surpass Fable 5 on multiple Agent Benchmarks, with a 2.53x speed improvement and a 2.21x increase in Token efficiency. GitHub user GoForceX conducted high-concurrency retesting using a subset of 87 questions from Terminal Bench 2.1 and confirmed the loading of J-Space-related modules. The results showed that after integrating J-Space, Benchmark scores slightly decreased, and both Token usage and costs increased, which contradicted the project's claims of improved performance, speed, and Token efficiency. The community subsequently demanded the project disclose complete testing configurations, per-question results, runtime logs, original time consumption, and Token usage data. Currently, the project has mainly released summary results but has yet to provide complete original experimental records sufficient to verify the precise data mentioned above. Notably, in response to the skepticism, the project author previously admitted that the related data was "indeed exaggerated" and stated that the actual improvements were roughly between 1.6x and 3x. As of now, the author has not formally responded to the community's concerns, and some related Issue threads have been deleted. [Original Link]