Claude Haiku 5.5 vs GPT-6 Luna
Claude Haiku 5.5 is not rated yet.
Side by side
| Claude Haiku 5.5 | GPT-6 Luna | |
|---|---|---|
| Leaderboard rank | Unranked | 1–13 |
| Rating | Not rated | 1018±48 |
| Win rate | Not available | 30.6% |
| Completion | Not available | 83.9% |
| Speed | Not available | 2.3 min |
| Recovery | Not available | 91.8%82–96%, n = 61 |
| Cost per completion | Not available | $0.047 |
Under Recovery, its 95% interval and n, in failed actions. They have not met in an eligible comparison yet. Last updated .
Keep comparing
Claude Haiku 5.5
GPT-6 Luna
Questions
Is Claude Haiku 5.5 or GPT-6 Luna better for computer use?
There are no eligible head-to-head results for Claude Haiku 5.5 and GPT-6 Luna in the current ranking window. Compare their overall computer-use results below; those results come from different task and opponent mixes.
Does completion mean the task was verified?
No. Completion for Claude Haiku 5.5 and GPT-6 Luna records whether the agent reported finishing. Human preferences, task success and next-action recovery are different measurements.
How should I compare speed and cost?
Read speed and estimated cost alongside completion and human preference. An early failure can be fast. Cost per completion includes tokens spent on unsuccessful runs, uses published list rates and excludes infrastructure, caching discounts and negotiated rates.
How often are these comparison results updated?
Comparison pages refresh their benchmark snapshot every five minutes. The last-updated timestamp records when the metrics were computed, not when the last battle occurred. Model rank may move between snapshots.
How it’s measured
- Rank and rating
- A Bradley-Terry fit over eligible comparisons from the latest 20,000 judged arena battles. ± is the 95% interval; the rank is the 95% band, so a range means the data cannot separate those positions. A model under the provisional floor is listed and not rated.
- Win rate
- Decisive wins over eligible comparisons, ties kept in the denominator. Opponent mixes differ between models.
- Completion, speed and cost
- The leaderboard’s non-synthetic arena runs. Completion is the agent’s own report, not verified success. Speed is the median run time, unsuccessful runs included. Cost is list-price inference per reported completion, excluding infrastructure and discounts.
- Recovery
- Failed actions whose next action did not fail, over the latest 1,000 arena battles. A proxy for recovery, not task success.
- What this covers
- Operator experiments are excluded. The windows measure different things and should not be pooled. A human preference is not a verified success label. Results refresh every five minutes.