Forecasting on AIBQ2
32.1MBSKimi
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| KimiModel capacity=Kimi2026.04 | 32.1 | — | — | |
| FlashModel capacity=Flash2026.04 | 32.3 | — | — | |
| BLF (no calibration)Calibration=none2026.04 | 34.8 | — | 14.6 | |
| BLF (nobelief)Belief state=none2026.04 | 35.1 | — | — | |
| BLF (zs)Search=zero-shot (no search)2026.04 | 36.7 | — | — | |
| BLF (medium thinking level)Thinking budget=medium2026.04 | 37.5 | — | — | |
| BLF +caltrials=mean of 5 trials, Calibration=Platt calibration (LOO CV)2026.04 | 39.5 | 73.8 | 13.6 | |
| SOTA (RL)2026.04 | 45.8 | — | — | |
| clairvoyantSearch cutoff=date to the present (sees future data)2026.04 | 71.7 | 89.3 | 5.8 |