FLY-EVAL++ - Single-Step Prediction - Field Validity: leaderboard

Metric: S1 D2 field score (%). Source: arxiv.org. 21 models tracked.

Top models

#ModelScore
1Grok 468
2Qwen 3 235B A22B63.8
3Qwen 3 32B63.7
4Qwen 2.5 32B63.6
5Qwen3-Next-80B63.4
6Claude Sonnet 4.561.9
7Llama 3.1 405B61.9
8Kimi K2 090561.8
9Gemini 2.5 Pro61.7
10DeepSeek V361.7
11DeepSeek R1 052861.7
12GPT-561.6
13GPT-4o61.6
14DeepSeek V3.161.5
15Kimi K2 (Thinking)61.5

Interactive version: theaggregate.ai/benchmark?slug=fly-eval-plus-plus-single-step-prediction-field-validity · How It Works · Data refreshed daily, snapshot 2026-09-19.