# Multimodal reasoning

GLM-4.1V-9B-Thinking is an open-source vision-language model based on the GLM-4-9B-0414 foundation model, focusing on improving the reasoning ability in complex tasks and supporting a 64k context length and 4K image resolution.

Kimi-VL-A3B-Thinking-2506 is an upgraded version of Kimi-VL-A3B-Thinking, with significant improvements in multimodal reasoning, visual perception and understanding, video scene processing, etc. It supports higher-resolution images and can achieve more intelligent thinking while consuming fewer tokens.

Magistral-Small-2506-Vision is an inference fine-tuned version based on Mistral Small 3.1 with GRPO training, an experimental checkpoint with visual capabilities.

Stockmark-2-VL-100B-beta is a Japanese-specific vision-language model with 100 billion parameters, equipped with chain-of-thought (CoT) reasoning ability and can be used for document reading and comprehension.

InternVL3 - 8B is an advanced multimodal large - language model with excellent multimodal perception and reasoning capabilities, capable of processing multimodal data such as images and videos.

InternVL3 - 1B is an advanced multimodal large language model that excels in multimodal perception, reasoning, and other abilities. It also expands multimodal capabilities such as tool use and GUI agent.

VisionReasoner-7B is an image-text-to-text model that adopts a decoupled architecture and consists of a reasoning model and a segmentation model. It can interpret user intentions and generate pixel-level masks.

Qwen3-8B is the latest large language model in the Qwen series. It has a variety of advanced features, supports multiple languages, and performs excellently in reasoning, instruction following, etc., bringing users a more intelligent and natural interaction experience.

Synthia-S1-27b is an advanced reasoning AI model developed by Tesslate AI, focusing on logical reasoning, coding, and role-playing tasks.

GGUF quantized version of Gemma 3 with 27B parameters, supporting image-text interaction tasks

Spec-Vision-V1 is a lightweight, state-of-the-art open-source multimodal model designed for deep integration of visual and textual data, supporting a 128K context length.

The Mulberry model is a step-by-step reasoning-based model trained on the Mulberry - 260K SFT dataset generated through collective knowledge search.

This is an open-source model under the Apache-2.0 license. Specific information needs to be supplemented.

This is an open-source model based on the Apache-2.0 license. Specific functionalities should be referenced in the actual model documentation