CarbonWatch

Chat services

What people call a chatbot: a question, a written answer, produced token by token. It is the most common type, and the one whose estimate is the most direct — the bytes arriving are the answer being written.

Billed on volume received, converted into tokens and then energy at 2 Wh per thousand tokens.

Service	Inference region	Intensity	500-token answer
Claude	United States	380 g/kWh	0.38 g · 500 tokens
ChatGPT / OpenAI	United States	380 g/kWh	0.38 g · 500 tokens
Grok / xAI	United States	380 g/kWh	0.38 g · 500 tokens
Meta AI	United States	380 g/kWh	0.38 g · 500 tokens
Mistral / Le Chat	France	50 g/kWh	0.05 g · 500 tokens
DeepSeek	China	580 g/kWh	0.58 g · 500 tokens
Microsoft Copilot	United States	380 g/kWh	0.38 g · 500 tokens
Poe	United States	380 g/kWh	0.38 g · 500 tokens
OpenRouter	World	480 g/kWh	0.48 g · 500 tokens
Groq	United States	380 g/kWh	0.38 g · 500 tokens
Together AI	United States	380 g/kWh	0.38 g · 500 tokens
Fireworks	United States	380 g/kWh	0.38 g · 500 tokens
Cohere	United States	380 g/kWh	0.38 g · 500 tokens
Hugging Face Inference	World	480 g/kWh	0.48 g · 500 tokens
Qwen / Tongyi	China	580 g/kWh	0.58 g · 500 tokens
Kimi / Moonshot	China	580 g/kWh	0.58 g · 500 tokens
Zhipu / GLM	China	580 g/kWh	0.58 g · 500 tokens
Ernie / Baidu	China	580 g/kWh	0.58 g · 500 tokens
Doubao / ByteDance	China	580 g/kWh	0.58 g · 500 tokens
Z.ai	China	580 g/kWh	0.58 g · 500 tokens
Character.AI	United States	380 g/kWh	0.38 g · 500 tokens
Pi / Inflection	United States	380 g/kWh	0.38 g · 500 tokens
AWS Bedrock (US)	United States	380 g/kWh	0.38 g · 500 tokens
AWS Bedrock (UE)	European Union	250 g/kWh	0.25 g · 500 tokens
DeepL	European Union	250 g/kWh	0.06 g · 500 tokens
Grammarly	United States	380 g/kWh	0.1 g · 500 tokens