VLAN - (b) BVI - Instruction Agreement: leaderboard

Full1931trajectorysplit; user-conditionedlanguageinstruction. SRlan requiresalluser-specificsemanticconstraints; IAS weightedstructuredsimilaritytocanonicaltarget. Seven releasedVLMs; exactInstructlabelsretained; noactionSRderivedmetricused. Metric: Native score (0–1). Source: https://arxiv.org/html/2611.18346v1

Metric: Native score (0–1). Source: arxiv.org. Saturation forecast: Estimated already saturated. 7 models tracked. Updated 17 Nov 2026.

Top models

#ModelScoreOverall rank
1Molmo2-8B0.781#1629
2InternVL3.5-8B0.774#1588
3Qwen 3 VL 8B Instruct0.767#1025
4Qwen 2.5 VL 3B Instruct0.728#2648
5Llama 3.2 11B Instruct0.723#5226
6Qwen 2.5 VL 7B Instruct0.719#1893
7Idefics3-8B-Llama30.576#5775

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=vlan-b-bvi-instruction-agreement · How It Works · Data refreshed daily, snapshot 2026-11-17.