Same standard of work. About 13× the energy.
Claude Haiku 5.5 scores higher than GPT-4o on Epoch's Capabilities Index, 158.98 against 128.73, across 58 benchmarks covering coding, agents, long context and writing, while using about 13× less energy per answer. Claude Haiku 5.5 has not faced human head-to-head voting, where GPT-4o carries 112,881 votes, worth weighing.
| Claude Haiku 5.5 | GPT-4o | |
|---|---|---|
| Energy per answer | 0.515 Wh | 6.92 Wh |
| Carbon released | 0.24 g CO₂ | 3.2 g CO₂ |
| Same carbon as driving | under a metre | 13 metres (42 feet) |
| Energy rank | 98 of 277 | 211 of 277 |
| Capability (58 benchmarks) | 159.0 | 128.7 |
| Human vote rating | - | 1300 |
| Votes cast | - | 112,881 |
| Confidence | Day-one estimate | Day-one estimate |
| Released | 2026-10-07 | 2024-08-06 |
Energy is calculated the same way for every model on this site, from public data and calibrated against laboratory measurement, accurate to roughly a factor of two to three, which is why we only ever state a difference of 5× or more. Capability is Epoch AI's Capabilities Index, one score combining 58 separate benchmarks; the vote rating comes from blind human comparisons. The full method is here.