XDomainBench: leaderboard
Large Language Models (LLMs) are increasingly deployed for knowledge synthesis, yet their capacity for compositional generalization in scientific knowledge remains under-characterized. Existing benchmarks primarily focus on single-turn restricted scenarios, failing to capture the capability boundaries exposed by real-world interactive scientific workflows. To address this, we introduce XDomainBenc
Metric: R (k=1, Deterministic) (self-reported). Source: benchmarklist.com. Saturation forecast: Around April 2027. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3-Next-80B | 63.8 | #312 |
| 2 | Mixtral 8x7B | 56.4 | #1143 |
| 3 | GPT-5.2 | 30.8 | #110 |
| 4 | Claude Haiku 4.5 | 30.6 | #280 |
| 5 | Claude Sonnet 4.5 | 30.4 | #142 |
| 6 | Gemma 2 2B (IT) | 27.3 | #1341 |
| 7 | Gemini 2.0 Flash | 27.2 | #342 |
| 8 | Qwen 2.5 14B | 27.2 | #725 |
| 9 | Gemini 2.5 Flash | 27.1 | #245 |
| 10 | Llama 3.2 3B | 26.8 | #1396 |
| 11 | Qwen 2.5 72B | 26.8 | #499 |
| 12 | Qwen 2.5 7B | 26.4 | #922 |
| 13 | Llama 3.1 8B | 26.2 | #1151 |
| 14 | GPT-5 Mini | 23.1 | #186 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=xdomainbench · How It Works · Data refreshed daily, snapshot 2026-10-20.