Environmental Impact Analysis on Observatory Standardized Request market-model catalog current
0.0056Inference Wh/requestMinistral 3B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Ministral 3BInterpretation=Low-end model illustrating the lower bound of the current catalog2026.03 | 0.0056 | 0.0002 | 0.0004 | |
| GPT-4o miniInterpretation=Lightweight hosted model with low per-request screening impacts2026.03 | 0.0155 | 0.006 | 0.0027 | |
| Llama 3.1 70BInterpretation=Large open-weight model with moderate inference estimates2026.03 | 0.1043 | 0.0402 | 0.16 | |
| GPT-5 miniInterpretation=Medium model representative of lighter hosted deployments2026.03 | 0.1706 | 0.0657 | 0.28 | |
| Gemini 2.5 ProInterpretation=Upper-mid-range model with markedly lower screening values than the top tier2026.03 | 0.3601 | 0.1387 | 1.26 | |
| GPT-5.2Interpretation=Frontier proprietary model illustrating the high end of current screening estimates2026.03 | 2.7897 | 1.074 | 101.76 | |
| Claude Opus 4.1Interpretation=Upper-end proprietary model with very large screening values for both inference and training2026.03 | 2.9922 | 1.152 | 125.63 |