Ecomputy

User profile picture

Efficiency Scoreboard

Tokens/kWh

1.2M

Inferences/kWh

8.5K

Cost/Inference

$0.0012

Carbon Emissions

120 kgCO2e

Carbon Mapper

A world map showing carbon intensity hotspots

Alerts

High Emissions

Model 'GPT-4-Turbo' in US-East-1 exceeds emission threshold.

Underutilized GPU

GPU usage for 'LLaMA-2-7B' is at 30% capacity.

Cost Spike

Inference costs for 'Claude 3' have increased by 40%.

Recommendations

Optimize Batch Size

Increase batch size for 'LLaMA-2-7B' to improve GPU utilization.

-15 kgCO2e

-$250/month

Shift Workload

Move 'GPT-4-Turbo' workload from US-East-1 to EU-West-3 (lower grid intensity).

-40 kgCO2e

-$50/month

Model Quantization

Apply 8-bit quantization to 'Claude 3' to reduce inference cost.

-5 kgCO2e

-$800/month