Generative Insight Anticipation on GiantsBench (test)
5.47Economics ScoreGIANTS-4B
Evaluation Results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GIANTS-4BJudge LM=gemini-3-pro2026.04 | 5.47 | — | 5.5 | 5.52 | 5.63 | 5.66 | 5.75 | 5.76 | 6.04 | 6.05 | 6.14 | 6.21 | 6.22 | 6.23 | 6.46 | 6.65 | 6.7 | 5.97 | |
| SFT-thinkJudge LM=gemini-3-pro2026.04 | 4.94 | — | 4.84 | 4.57 | 4.8 | 4.81 | 5.31 | 4.87 | 5.13 | 5.33 | 5.3 | 5.11 | 5.08 | 5.02 | 5.51 | 5.69 | 5.42 | 5.05 | |
| SFTJudge LM=gemini-3-pro2026.04 | 4.78 | — | 4.75 | 4.59 | 4.72 | 4.73 | 5.15 | 4.97 | 5.07 | 5.31 | 5.27 | 5.06 | 5.02 | 5.12 | 5.38 | 5.43 | 5.58 | 5.01 | |
| gemini-2.5-proJudge LM=gemini-3-pro2026.04 | 4.75 | — | 4.45 | 4.11 | 4.67 | 4.76 | 5.04 | 4.64 | 4.57 | 4.4 | 4.95 | 4.55 | 4.74 | 4.52 | 5.5 | 4.94 | 4.57 | 4.65 | |
| BaseJudge LM=gemini-3-pro2026.04 | 4.66 | — | 4.5 | 4.28 | 4.54 | 4.64 | 4.82 | 4.77 | 4.62 | 4.87 | 4.82 | 4.74 | 4.83 | 4.95 | 5.45 | 5.17 | 5.16 | 4.75 | |
| gemini-3-proJudge LM=gemini-3-pro2026.04 | 4.5 | — | 4.14 | 4.14 | 4.58 | 4.58 | 4.45 | 4.56 | 4.41 | 3.82 | 4.65 | 4.43 | 4.55 | 4.01 | 5.68 | 4.41 | 4.19 | 4.43 |