GPT-5.6 Luna
OpenAI · Codex CLI · Queued
Common-task access check finished; full public pass queued. 2 of 120 current-version public tasks completed. Recorded reasoning effort: max. One attempt per task; provisional pilot grades.
| Model / run status | Tasks completed | Tasks passed | Pass rate | Cost / task | Cost / solved task |
|---|---|---|---|---|---|
| GPT-5.6 Luna OpenAI · Queued | 2 / 120 Codex CLI · max effort | 2 / 2 | 100.0% | $0.0068 | $0.0068 |
1 models shown. A dash means no valid measurement, unavailable pricing, or no solved tasks. Costs are list-price equivalents over completed, current-version tasks. Network failures and transport errors are excluded. Read limitations.
Results by benchmark suite
| Model | Reason 24 tasks | Datasheet 12 tasks | Select 12 tasks | Library 12 tasks | Schematic 12 tasks | Review 16 tasks | ECO 8 tasks | Layout 12 tasks | Board 6 tasks | Debug 6 tasks |
|---|---|---|---|---|---|---|---|---|---|---|
| GPT-5.6 Luna | 2/2 | — | — | — | — | — | — | — | — | — |
Native board outputs
No board previews match these filters. Some attempts produced no render or native board.
0 of 0 previews shown. Rendered from recorded native KiCad outputs. Missing 3D component models appear as pads. These are digital designs, not fabricated or bench-tested boards.
Interpretation and cost basis
No Season 1 rank is assigned. Partial runs have different task coverage, and token costs are list-price equivalents of subscription usage. Read all pilot limitations.
CLI model: gpt-5.6-luna · Last recorded attempt: Oct 15, 2026