Run #105
pilot running claude-opus-4-6.
82%95% interval 74.2–88.5%Ranks 1–4 of 8 on this board
Details
What the site recorded for this run. This address keeps pointing at it.
- Entry
pilot-real-claude-opus-4-6- Agent
- pilot
- Model
claude-opus-4-6- Date
- 28 Mar 2026
- hbench release
- converted from another tool's job
- hbench results epoch
- 0.2
Cite this run
pilot (pilot, entry pilot-real-claude-opus-4-6), Terminal-Bench 2.0 on Claude Opus 4.6 (5 attempts, open network): 82% (95% interval 74.2–88.5%), run #105, 28 Mar 2026. https://nodeq.tech/runs/105
365 / 445trials solved
$0.67cost per solved trial
$244total cost
7.5 minmedian agent time
572k/9.6Mtokens in / out
By category
The share of trials solved in each kind of task.
- software-engineering85 / 130
- system-administration24 / 45
- data-science39 / 40
- scientific-computing39 / 40
- security39 / 40
- debugging25 / 25
- file-operations25 / 25
- data-processing20 / 20
- mathematics20 / 20
- model-training10 / 20
- machine-learning15 / 15
- data-querying5 / 5
- games5 / 5
- optimization5 / 5
- personal-assistant4 / 5
- video-processing5 / 5