InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency

TL;DR

InternVL3.5 introduces innovative training and deployment strategies for open-source multimodal models, significantly improving reasoning, efficiency, and versatility, and achieving state-of-the-art results in various tasks.

Contribution

The paper presents Cascade RL, ViR, and DvD strategies that enhance reasoning, optimize efficiency, and enable new capabilities in open-source multimodal models.

Findings

Up to 16% improvement in reasoning performance

4.05x faster inference speed

State-of-the-art results on multiple benchmarks

Abstract

We introduce InternVL 3.5, a new family of open-source multimodal models that significantly advances versatility, reasoning capability, and inference efficiency along the InternVL series. A key innovation is the Cascade Reinforcement Learning (Cascade RL) framework, which enhances reasoning through a two-stage process: offline RL for stable convergence and online RL for refined alignment. This coarse-to-fine training strategy leads to substantial improvements on downstream reasoning tasks, e.g., MMMU and MathVista. To optimize efficiency, we propose a Visual Resolution Router (ViR) that dynamically adjusts the resolution of visual tokens without compromising performance. Coupled with ViR, our Decoupled Vision-Language Deployment (DvD) strategy separates the vision encoder and language model across different GPUs, effectively balancing computational load. These contributions collectively enable InternVL3.5 to achieve up to a +16.0\% gain in overall reasoning performance and a 4.05$\times$ inference speedup compared to its predecessor, i.e., InternVL3. In addition, InternVL3.5 supports novel capabilities such as GUI interaction and embodied agency. Notably, our largest model, i.e., InternVL3.5-241B-A28B, attains state-of-the-art results among open-source MLLMs across general multimodal, reasoning, text, and agentic tasks -- narrowing the performance gap with leading commercial models like GPT-5. All models and code are publicly released.