P

Prompt-per-Mile

Offset Calculator

Tune your ride

Slide to see the prompts your green transit offsets.

10 mi
0250500

Transit mode

E_train 0.075 lbs CO₂/mi

Prompt complexity

0
010,00020,000

Slide to weigh your own AI usage against these miles.

Live offset

3.25 lbs

CO₂ saved · 1.47 kg

12,745

text prompts offset

Your prompt balance

0 prompts × 0.3 Wh = 0.00 lb CO₂

Covered — you still have 12,745 text prompts in credit, and 3.25 lb of CO₂ still banked.

You earned 12,745 text prompts or 2,549 complex debugging queries!

Every green mile quietly powers the next prompt.

0.40

lb CO₂/mi · car

0.075

lb CO₂/mi · passenger train

0.00085

lb CO₂/Wh · grid

P = (M × (E_car − E_train)) ÷ (e_prompt × C_grid) · currently ( 10 × (0.40 − 0.075) ) ÷ (0.3 × 0.00085)

Ask the trip advisor

Describe your trip — AI picks the greenest way there

E.g. "Tuesday morning from downtown Phoenix to Tempe campus with a backpack, about 20 minutes of walking is fine."

Reference sources

Where these numbers come from

  1. 1.Greenhouse Gas Emissions from a Typical Passenger Vehicle — U.S. Environmental Protection Agency. Source of the 404 g CO₂ per mile average for a gasoline car, rounded here to 0.40 lb CO₂/mile as the driving baseline (E_car).
  2. 2.National Transit Database — U.S. Federal Transit Administration. Passenger-mile, fuel, and energy reporting behind the per-passenger-mile factors used for passenger rail, commuter rail, and bus modes.
  3. 3.Transportation Energy Data Book — Oak Ridge National Laboratory / U.S. Department of Energy. Energy intensity and average occupancy data used to approximate per-passenger emissions for buses, shuttle vans, and light-duty vehicles.
  4. 4.eGRID — Emissions & Generation Resource Integrated Database — U.S. Environmental Protection Agency. Basis for the grid carbon intensity constant C_grid = 0.00085 lb CO₂ per watt-hour of electricity, and for electric-vehicle emissions estimates.
  5. 5.How much energy does ChatGPT use? — Epoch AI (Josh You, 2025). Source of the ≈0.3 Wh per standard text prompt estimate; the 1.5 Wh deep reasoning / coding tier scales that figure for longer inputs and outputs.
  6. 6.Measuring the environmental impact of delivering AI at Google scale — Google Cloud. Independent per-prompt measurement (≈0.24 Wh median text prompt) used as a cross-check on the prompt energy tiers.

Transit and rideshare factors are per-passenger-mile averages that vary widely with occupancy, vehicle age, and local grid mix. Values for LNG buses, shuttle vans, electric autos, and robotaxis are teaching approximations derived from the transit and grid sources above, not manufacturer-certified figures.