GOOGLE

Gemini 4 Neon Claims the #1 Spot on the Models Engineering Index

Google's latest model tops the composite software engineering benchmark, outperforming rivals in practical coding tasks.


The Models Engineering Index (MEI)—which evaluates language models on practical engineering benchmarks adjusted for difficulty, missing results, and token costs—has crowned a new leader. Google's 'Gemini 4 Neon' captured the #1 position with a composite index score of 68.1.

Anthropic's Fable 5.2 secured the second spot with 66.3 points, followed by OpenAI's 6.2 Astra at 62.0. The MEI benchmark standardizes performance across ten distinct coding evaluations while scaling cost and token usage relative to industry baselines, offering engineering teams a rigorous metric for practical software development efficiency.