GreenPT GreenL (Mistral Small 3.2 24B)

GreenPT

sustainable_balanced

Capabilities

VisionToolsCodeStructured OutputMultilingual

Strengths

100% renewable energy inference
Heat recovery — server heat heats buildings
Real-time energy metrics (mWh/100 tokens)
EU-hosted GDPR compliant
No external API calls
Open source model
Vision support via Mistral Small 3.2
OCR and Scraper APIs for document processing

Weaknesses

Smaller model (24B) — not frontier quality
Subscription required for API
Training not on green infrastructure

Pricing

Input / 1M tokens

$0.25

Output / 1M tokens

$0.8

Context window

131k

Transparency

Open weights

10.0 / 10

Open training data

3.0 / 10

Open methodology

5.0 / 10

Licence openness

10.0 / 10

Provider disclosure

8.0 / 10

FMTI company score

N/A

Composite:7.2 / 10

Not scored by FMTI (too small). Publishes real-time energy metrics. Fully open source models. Transparent about limitations.

Sustainability

Inference energy

10.0 / 10

Training footprint

3.0 / 10

Provider infrastructure

10.0 / 10

Composite:7.7 / 10

100% renewable inference. Heat recovery for community heating. Published mWh/100 tokens. PUE 1.37 (industry avg 1.57). WUE 0.067 (industry avg 1.8). 20-30% compute reduction through compression. Training not on green infra (acknowledged).

Task Fitness

question answering

67%

code

72%

mathematics

69%

business communication

58%

analysis

65%

extraction

75%

long-form generation

68%

research

45%

reasoning

65%

summarisation

72%

translation

75%

conversation

70%