Google Gemini 4 Carbon ARC-AGI Benchmark Results

Google evaluated its Gemini 4 Carbon model on the ARC-AGI benchmark, achieving verified scores of 99.1% on ARC-AGI-3 using a provider adapter and 65.1% using a standard harness. The results show that Carbon beat its predecessor Argon by 18 percentage points on ARC-AGI-3 in the standard harness. Open article →