PRISM-VLM - Instruction Following: leaderboard
Metric: Per-axis mean score (%; mean compliance with 1-3 verifiable IFEval-style formatting constraints appended to each item; over PRISM-VLM's 6,238 items recycled from 15 public VLM benchmarks (five seeds of 100 items per benchmark); GPT-5 (low effort) synthesizes the perturbations and grades the open-ended axes; each model at the lowest reasoning effort its provider exposes, temperature 0). Source: arxiv.org. Saturation forecast: Around December 2026. 42 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 Mini (Minimal) | 87.5 |
| 2 | GPT-5.4 Mini | 87.1 |
| 3 | GPT-5.4 Nano | 86 |
| 4 | GPT-4o Mini | 84.3 |
| 5 | GPT-5 Nano (Minimal) | 83.7 |
| 6 | Qwen 3 VL 8B Instruct | 82.3 |
| 7 | GPT-4.1 Nano | 81.7 |
| 8 | Gemini 2.5 Flash (Non-reasoning) | 81.4 |
| 9 | Gemini 3 Flash (Minimal) | 81 |
| 10 | GPT-4.1 Mini | 80.5 |
| 11 | Molmo2-8B | 79.8 |
| 12 | Gemini 2.0 Flash | 79.1 |
| 13 | Qwen 3 VL 4B Instruct | 78.9 |
| 14 | Claude 3 Haiku | 77.8 |
| 15 | Nova 2 Lite | 77.4 |
Interactive version: theaggregate.ai/benchmark?slug=prism-vlm-instruction-following · How It Works · Data refreshed daily, snapshot 2026-09-26.