Gemini 4 Carbon Frontier AI Model Falls Slightly Behind Claude Opus 5.5 in Agentic Capabilities, but Not in Cost Efficiency
According to SemiAnalysis' GDPx benchmark, Gemini 4 Carbon achieved 10.9 percentage points in agentic capabilities, slightly below Claude Opus 5.5's 12 percentage points. However, Gemini 4 Carbon completed tasks at $16.29 per task using 594,000 output tokens, compared to Claude Opus 5.5's $51.41 per task with 548,000 output tokens, demonstrating significantly lower operational costs.
Key points
- Gemini 4 Carbon scored 10.9 vs Claude Opus 5.5's 12 in GDPx agentic benchmark
- Google model is 3x more cost-efficient at $16.29 vs $51.41 per task
- Performance gap minimal but cost advantage significant for deployment
Summary and key points written by AI from the headline and a short excerpt, so they can be incomplete or wrong. Gadgets 360 did the reporting; read the original for the full story.
Read the full story at Gadgets 360
