Vision & Multimodal Aug 2025 · #3 most cited · 1.2K citations

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

Weiyun Wang et al.

arXiv:2508.18265

In short

InternVL3.5 is an open family of multimodal models trained with a two-stage “cascade” of offline then online reinforcement learning to improve reasoning. A router that adapts image resolution and a deployment that splits the vision encoder and language model across GPUs make it about four times faster than its predecessor.

Why it matters

It narrows the gap between open and commercial multimodal models on reasoning and agent tasks.

Read first

The 3 Field Guide ideas this paper leans on.

Starting from scratch? The full route 23 ideas · basics first
  1. Machine Learning ✓ understood

    Building systems that learn patterns from data instead of following hand-written rules, getting better at a task as they see more examples.

  2. Neural Network ✓ understood

    A model built from layers of simple units, each taking a weighted sum of its inputs and applying a nonlinearity; training adjusts the weights so the whole stack maps inputs to useful outputs.

  3. Deep Learning ✓ understood

    A subset of machine learning that uses neural networks with multiple layers (deep neural networks) to learn hierarchical representations of data.

  4. Computer Vision ✓ understood

    The field of AI that gets computers to extract meaning from images and video: what is in them, where it is, and how it moves.

  5. Natural Language Processing ✓ understood

    The field of AI that lets computers read, interpret, translate and generate human language, from spam filters and search to chatbots.

  6. Token ✓ understood

    The unit of text a language model reads and writes: a word, part of a word, or a character, drawn from the model's fixed vocabulary.

  7. Tokenization ✓ understood

    Splitting text into tokens, usually subword pieces, and mapping each to an integer ID so a language model can process it.

  8. Language Modeling ✓ understood

    Assigning probabilities to sequences of text, in practice by predicting each next token from the ones before it; the objective behind large language models.

  9. Dataset ✓ understood

    A collection of data examples used for training, validating, or testing machine learning models.

  10. Training ✓ understood

    The process of fitting a model to data by repeatedly measuring how wrong its outputs are and adjusting its parameters to reduce that error.

  11. Unsupervised Learning ✓ understood

    Learning from unlabeled data to discover hidden patterns, structures, or relationships without explicit target outputs.

  12. Self-Supervised Learning ✓ understood

    Training on unlabeled data by making the data supply its own answers: hide or alter part of an input and have the model predict or match it.

  13. Pre-training ✓ understood

    Training a model on a large dataset (often self-supervised) before fine-tuning on specific tasks, enabling transfer learning.

  14. Feature ✓ understood

    A single measurable property of an example, such as a house's floor area or how many links an email contains, used as an input to a model.

  15. Representation Learning ✓ understood

    Letting a model learn its own internal description of the data, such as embeddings or layered visual features, instead of relying on features designed by hand.

  16. Embedding ✓ understood

    A list of numbers (a vector) that represents a word, sentence, image or other item, learned so that similar items end up close together.

  17. Attention Mechanism ✓ understood

    A technique that lets a neural network weigh every part of its input when producing each output, focusing on the parts most relevant at that step.

  18. Transformer ✓ understood

    A neural network architecture, introduced in 2017, built from stacked self-attention and feed-forward layers; the basis of nearly every modern large language model.

  19. Large Language Model ✓ understood

    A neural network, almost always a transformer, trained on vast amounts of text to predict the next token, which lets it write, answer, summarize and follow instructions.

  20. Multimodal Model · read first ✓ understood

    Models processing multiple data types (text, images, audio) jointly, like GPT-4V, Gemini, or CLIP.

  21. Reinforcement Learning · read first ✓ understood

    Learning through interaction with an environment, receiving rewards or penalties to learn optimal behavior policies.

  22. Inference ✓ understood

    Running a trained model on new inputs to get predictions, with its weights frozen: the stage of a model's life that users actually interact with.

  23. Inference Latency · read first ✓ understood

    The time delay between submitting input and receiving output from a deployed model, critical for real-time applications.

In the frontier

Rank
#3 of 100
Citations
1.2K
as of Aug 9, 2026
Published
Aug 2025

Topics: Vision-language models , RL for reasoning , Efficiency and serving

Selection: 1kpapers.com by Together AI, most-cited as of Aug 9, 2026

Nearby papers

Summary in our own words; read the paper for the details. ← All papers