The UK AI Security Institute has adopted EvalEval's infrastructure to openly share evaluation results, advancing reproducibility in AI model benchmarking. AISI and the EvalEval Coalition, collaborating on the "Every Eval Ever" (EEE) schema, are releasing verified evaluation data for five benchmarks tested on six frontier models including Claude Opus 4/4.5/4.6 and GPT-5/5.2/5.4. The initiative addresses a critical gap in the AI industry: evaluation results are typically reported across disparate platforms and formats without sufficient information for reproduction or verification. The partnership puts shared infrastructure into practice by making AISI's evaluation methodology and results publicly available through Evaluation Cards. This transparency is particularly important because reproducing evaluations independently can be prohibitively expensive, making openly published reference points essential for researchers and practitioners to understand how evaluation protocol choices—such as feedback mechanisms and token budgets—influence reported performance. The collaboration builds on AISI's previous work optimizing evaluation efficiency and statistical rigor. As AI deployment accelerates, standardized evaluation reporting offers broader benefits for meta-research and policy analysis. By adopting common schemas and sharing detailed evaluation context, the initiative enables researchers to compare model performance across different setups and identify when similar benchmark scores actually reflect different underlying capabilities. AISI and EvalEval are calling for broader adoption, encouraging model developers to report verified results and evaluation researchers to use the standardized schema.