Text / Chat

DeepSeek-V3.1 Environmental Impact

HeavyEstimated

Hybrid V3+R1 model — 671B params, 37B active

Architecture
Mixture-of-Experts (hybrid V3 + R1)
Parameters
671B
Context
128,000 tokens
Provider
DeepSeek
5.0 Wh
Energy per query
2.8 g
CO₂ per query
19 mL
Water per query
17x more than
vs Google search

Energy per query

5.0 Wh

17x more than a Google search (0.3 Wh)

CO2 per query

2.8 g

China grid (550 gCO₂/kWh)

Water per query

19 mL

~54 queries to fill 1 litre

Processing location

DeepSeek Cloud (China)

Provider

DeepSeek

Category

Text / Chat

Grid carbon intensity

550 g CO2/kWh (30% renewable)

How does DeepSeek-V3.1 compare?

Ranked #113 of 152 models by energy per query

0 Wh2 Wh4 Wh6 Wh8 WhLLaMA 3.2 1BGemini 1.5 ProGPT-4.1 NanoDeepSeek-V3.1Google search (0.3 Wh)

Detailed Breakdown

Energy Consumption

DeepSeek-V3.1 combines the V3 chat model with R1 reasoning capabilities in a single model. At ~5.0 Wh per query for standard chat (non-reasoning), it activates only 37B of 671B parameters. Reasoning mode activates more experts and consumes significantly more energy.

Power Source & Carbon

Hosted on DeepSeek's Chinese infrastructure. Open-source and widely self-hosted globally.

Water Usage

At ~18.5 mL per query on Chinese infrastructure.

About DeepSeek-V3.1

DeepSeek-V3.1 is a 671B-parameter text and chat model from DeepSeek, released August 1, 2025. Hybrid V3+R1 model — 671B params, 37B active. At 5.0 Wh per query, it uses 17x the energy of a Google search. It runs on a Mixture-of-Experts (hybrid V3 + R1) architecture.

These figures are estimates derived from hardware specifications and API benchmarks — DeepSeek has not published official energy data for DeepSeek-V3.1. Actual consumption may vary significantly depending on batching, quantisation, and infrastructure optimisations that we cannot observe from outside.

DeepSeek-V3.1 in Context

100%
potential savings

The efficiency alternative

Gemini Nano performs the same type of task using just 0.01 Wh per query — 100% less energy than DeepSeek-V3.1. For a user sending 25 queries per day, switching would save 45.5 kWh per year.

45.6 kWh
per year

Your yearly DeepSeek-V3.1 footprint

At 25 queries per day, your annual DeepSeek-V3.1 usage consumes 45.6 kWh — roughly what a fridge uses in a month. That produces 25.1 kg of CO₂.

Key Insights

Open-source weights — can be self-hosted on infrastructure you control

What does your DeepSeek-V3.1 usage cost the planet?

Use our calculator to estimate your personal environmental footprint based on how often you use DeepSeek-V3.1.

Calculate My Compute

Frequently Asked Questions

How much energy does DeepSeek-V3.1 use per query?

Each DeepSeek-V3.1 query consumes approximately 5.0 Wh of energy. This is 17x more than a traditional Google search (~0.3 Wh).

What is DeepSeek-V3.1's carbon footprint?

Based on the carbon intensity of DeepSeek Cloud (China), each query produces approximately 2.8 g of CO2. The grid in this region has a carbon intensity of 550 g CO2/kWh with 30% renewable energy.

How much water does DeepSeek-V3.1 use?

Each query consumes approximately 19 mL of water, primarily used for cooling the data centers that process the request.

How does DeepSeek-V3.1 compare to a Google search?

A DeepSeek-V3.1 query uses 17x more than a Google search in terms of energy. A Google search uses approximately 0.3 Wh, while DeepSeek-V3.1 uses 5.0 Wh.

Technical Details

Architecture

Mixture-of-Experts (hybrid V3 + R1)

Parameters

671B

Context window

128,000 tokens

Release date

2025-08-01

Open source

Yes

Training data cutoff

2025-07