Impact & sustainability
Energy per run, model green scores, and forecasted footprint.
Last 30 days
2.31 kWh
≈ 972 g CO₂e · 12,480 runs · 4.82M tokens
Efficiency tips
- Route summarization to the smallest capable model (save ~62%).
- Enable semantic cache for repeated prompt classes (~28%).
- Batch background runs into off-peak hours where region carbon-intensity dips.
Region intensity
EU-North
42 gCO₂/kWh
US-West
210 gCO₂/kWh
APAC-SG
320 gCO₂/kWh
EU-Central
180 gCO₂/kWh
Model green scores
openai/gpt-5.5-mini0.18 Wh / 1k tokAgoogle/gemini-3-pro0.31 Wh / 1k tokAopenai/gpt-5.50.72 Wh / 1k tokBanthropic/claude-4.5-sonnet0.94 Wh / 1k tokBopenai/gpt-5.5-pro2.10 Wh / 1k tokC