GPT-6 Luna
Task results
Task catalog ↗| Task | Hardware | Agent | Provider | Trajectory | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| agentic-search | CPU | Codex | OpenRouter | 1.0 | 2h 0min | 69,569,763 | 295,470 | 69,124,471 | $1.41 | Trace → |
| agentic-search-vectorless | CPU | Codex | OpenRouter | 0.0 | 1h 10min | 36,321,762 | 229,594 | 36,019,370 | $0.60 | Trace → |
| biomedical-graph-retrieval | CPU | Codex | OpenRouter | 10.0 | 44min | 11,377,906 | 177,295 | 11,146,275 | $0.23 | Trace → |
| budgeted-bimetric-retrieval | CPU | Codex | OpenRouter | 18.2 | 36min | 7,577,611 | 101,573 | 7,437,861 | $0.14 | Trace → |
| code-context-retrieval | CPU | Codex | OpenRouter | 16.7 | 1h 35min | 51,839,738 | 353,680 | 51,348,852 | $1.23 | Trace → |
| evolving-personal-memory | CPU | Codex | OpenRouter | 32.8 | 45min | 15,087,765 | 191,631 | 14,836,036 | $0.28 | Trace → |
| filtered-ann-retrieval | CPU | Codex | OpenRouter | 0.0 | 55min | 9,514,295 | 160,940 | 9,317,119 | $0.20 | Trace → |
| financial-report-qa | CPU | Codex | OpenRouter | 10.6 | 54min | 14,524,405 | 171,722 | 14,286,415 | $0.26 | Trace → |
| icsi-meeting-memory | CPU | Codex | OpenRouter | 0.0 | 1h 37min | 43,594,608 | 306,495 | 43,104,028 | $1.06 | Trace → |
| incremental-vector-search | CPU | Codex | OpenRouter | 0.0 | 2h 0min | 52,081,738 | 211,774 | 51,787,046 | $0.76 | Trace → |
| join-aware-table-retrieval | CPU | Codex | OpenRouter | 40.0 | 37min | 10,151,616 | 150,772 | 9,922,632 | $0.20 | Trace → |
| large-scale-formula-retrieval | CPU | Codex | OpenRouter | 16.7 | 2h 0min | 13,206,754 | 180,318 | 12,987,387 | $0.25 | Trace → |
| long-document-reranking | CPU | Codex | OpenRouter | 33.0 | 1h 32min | 23,608,663 | 197,538 | 23,336,756 | $0.37 | Trace → |
| long-pdf-evidence | CPU | Codex | OpenRouter | 33.3 | 1h 22min | 25,919,795 | 273,195 | 25,580,157 | $0.59 | Trace → |
| long-pdf-evidence-vectorless | CPU | Codex | OpenRouter | 8.3 | 1h 9min | 24,541,032 | 246,671 | 24,138,950 | $0.59 | Trace → |
| memory-constrained-dense-retrieval | CPU | Codex | OpenRouter | 83.0 | 40min | 15,760,687 | 216,783 | 15,493,549 | $0.30 | Trace → |
| narrative-clue-search | CPU | Codex | OpenRouter | 6.2 | 2h 0min | 39,224,828 | 321,145 | 38,358,508 | $1.05 | Trace → |
| negative-constraint-retrieval | CPU | Codex | OpenRouter | 0.0 | 40min | 11,078,228 | 166,738 | 10,864,413 | $0.22 | Trace → |
| paper-metadata-retrieval | CPU | Codex | OpenRouter | 0.0 | 1h 4min | 20,783,784 | 218,731 | 20,507,870 | $0.37 | Trace → |
| reasoning-query-rewriting | CPU | Codex | OpenRouter | 30.0 | 2h 0min | 19,348,336 | 188,140 | 19,050,943 | $0.32 | Trace → |
| rust-bm25-search | CPU | Codex | OpenRouter | 24.1 | 33min | 7,251,291 | 128,589 | 7,018,201 | $0.16 | Trace → |
| scientific-paper-qa | CPU | Codex | OpenRouter | 20.0 | 2h 0min | 58,004,541 | 237,577 | 57,333,274 | $1.13 | Trace → |
| sparse-retrieval-pruning | CPU | Codex | OpenRouter | 0.0 | 49min | 14,483,338 | 177,938 | 14,255,623 | $0.26 | Trace → |
| visual-document-qa | CPU | Codex | OpenRouter | 17.8 | 48min | 17,102,874 | 189,670 | 16,725,088 | $0.31 | Trace → |
| budgeted-candidate-reranking | GPU | Codex | OpenRouter | 7.3 | 36min | 10,478,347 | 85,487 | 10,312,377 | $0.17 | Trace → |
| chunqiu-temporal-retrieval | GPU | Codex | OpenRouter | 1.7 | 49min | 13,553,651 | 184,124 | 13,278,544 | $0.28 | Trace → |
| code-embedding-finetuning | GPU | Codex | OpenRouter | 5.0 | 2h 47min | 19,197,198 | 103,161 | 18,996,060 | $0.27 | Trace → |
| code-reranker-finetuning | GPU | Codex | OpenRouter | 40.9 | 3h 7min | 13,884,690 | 91,089 | 13,725,215 | $0.20 | Trace → |
| composed-image-search | GPU | Codex | OpenRouter | 36.3 | 49min | 6,697,157 | 142,581 | 6,469,145 | $0.16 | Trace → |
| constrained-molecular-search | GPU | Codex | OpenRouter | 15.5 | 2h 40min | 55,719,386 | 200,567 | 55,410,774 | $0.98 | Trace → |
| decision-model-training | GPU | Codex | OpenRouter | 38.3 | 3h 31min | 21,749,702 | 165,556 | 21,531,216 | $0.33 | Trace → |
| document-page-search | GPU | Codex | OpenRouter | 7.0 | 1h 3min | 16,534,852 | 134,622 | 16,319,010 | $0.26 | Trace → |
| instruction-following-embedding | GPU | Codex | OpenRouter | 0.0 | 50min | 11,116,341 | 93,217 | 10,966,393 | $0.18 | Trace → |
| khmer-embedding-training | GPU | Codex | OpenRouter | 14.6 | 1h 20min | 9,512,541 | 119,096 | 9,343,679 | $0.17 | Trace → |
| layout-page-search | GPU | Codex | OpenRouter | 18.8 | 1h 4min | 14,568,292 | 104,329 | 14,393,597 | $0.22 | Trace → |
| legal-stance-retrieval | GPU | Codex | OpenRouter | 9.9 | 32min | 5,445,518 | 102,185 | 5,307,722 | $0.12 | Trace → |
| lightweight-query-retrieval | GPU | Codex | OpenRouter | 0.0 | 3h 18min | 129,909,430 | 487,875 | 129,306,297 | $2.77 | Trace → |
| methodology-inspiration-retrieval | GPU | Codex | OpenRouter | 3.4 | 45min | 13,552,387 | 145,519 | 13,357,183 | $0.23 | Trace → |
| multilingual-vector-postprocessing | GPU | Codex | OpenRouter | 26.4 | 1h 4min | 16,088,069 | 142,560 | 15,888,705 | $0.26 | Trace → |
| protein-text-bidirectional-retrieval | GPU | Codex | OpenRouter | 16.7 | 56min | 9,773,361 | 76,697 | 9,618,692 | $0.15 | Trace → |
| quantity-aware-retrieval | GPU | Codex | OpenRouter | 0.0 | 45min | 12,673,530 | 136,920 | 12,491,690 | $0.22 | Trace → |
| query-encoder-alignment | GPU | Codex | OpenRouter | 28.9 | 46min | 11,669,767 | 143,052 | 11,453,733 | $0.21 | Trace → |
| same-place-image-search | GPU | Codex | OpenRouter | 57.6 | 1h 2min | 9,296,237 | 113,042 | 9,131,007 | $0.17 | Trace → |
| sketch-text-search | GPU | Codex | OpenRouter | 3.5 | 1h 24min | 21,040,279 | 157,333 | 20,783,360 | $0.32 | Trace → |
| spoken-query-search | GPU | Codex | OpenRouter | 13.8 | 2h 0min | 18,913,018 | 190,681 | 18,426,150 | $0.34 | Trace → |
| training-error-tracing | GPU | Codex | OpenRouter | 0.0 | 24min | 7,523,401 | 93,179 | 7,356,046 | $0.14 | Trace → |
| video-moment-search | GPU | Codex | OpenRouter | 22.8 | 24min | 3,774,714 | 108,385 | 3,622,157 | $0.11 | Trace → |
No matching tasks. Try a different search.
47 selected evaluations · valid zero scores included
Score: task reward × 100. Tokens: coding usage; cached tokens are included in input.
Cost is estimated from recorded agent tokens at official API rates in USD (2026-11-01; DeepSeek off-peak pricing).