Continual Learning Bench - Gain: leaderboard
The learning column of Continual Learning Bench: the same run's reward minus the reward of the identical system reset between instances, so it isolates what was gained from experience rather than what the model brought in. Every system on the board comes out positive, and the best gain is about a quarter of the task ceiling. It is a difference score, not a level.
Metric: Agg. Gain (%). Source: continual-learning-bench.com. Saturation forecast: Around 2030. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | 24.14 | #93 |
| 2 | GPT-5.4 | 22.37 | #84 |
| 3 | Claude Opus 4.7 | 19.45 | #51 |
| 4 | Gemini 3 Flash | 15.5 | #102 |
| 5 | Gemini 3.1 Pro (Preview) | 8.1 | #59 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=continual-learning-bench-gain · How It Works · Data refreshed daily, snapshot 2026-10-24.