V-FiLLM - Simplified Statements (Multi-Turn): leaderboard
V-FiLLM (arXiv 2608.11047v1, Table 2): verified financial reasoning, where each question is generated from an executable computation tree over statement values so its answer is deterministic, on 250 mixed-depth questions. Accuracy with simplified statements when each question is split into sequential sub-queries, one expression-tree node per turn, percent. Higher is better.
Metric: Accuracy (%; sequential sub-queries). Source: arxiv.org. Saturation forecast: Estimated already saturated. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.7 Plus | 98 | #98 |
| 2 | GPT-OSS-120B | 97.6 | #346 |
| 3 | Gemma 4 31B (IT) | 96.4 | #178 |
| 4 | DeepSeek V4 Flash (0731) | 88 | #84 |
| 5 | Llama 3.3 70B Instruct | 86.8 | #541 |
| 6 | Qwen 3.5 9B | 86.8 | #391 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=v-fillm-simplified-statements-multi-turn · How It Works · Data refreshed daily, snapshot 2026-11-05.