XCR-Bench - CSI Identification: leaderboard

Metric: Soft F1 (%; normalised Levenshtein similarity of predicted and gold culture-specific item spans under one-to-one alignment). Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.

Top models

#ModelScore
1Claude 3.7 Sonnet60.47
2Qwen 2.5 Max59.75
3GPT-4o57.74
4OLMo 2 32B57.71
5Gemini 2.0 Flash57.15
6DeepSeek R1 Distill Qwen 32B55.14
7Llama 3.3 70B Instruct48.76

Interactive version: theaggregate.ai/benchmark?slug=xcr-bench-csi-identification · How It Works · Data refreshed daily, snapshot 2026-09-25.