AIAny
AI Model2026
Icon for item

Qwythos-27B-v1

27B multimodal reasoning model built on Qwen3.5-27B that preserves the base model's native multi-token-prediction head, full vision tower, and a 1,048,576-token YaRN context window. Designed for agentic tool use, long-context reasoning, and research deployments; released under Apache-2.0.

Visit Website

Introduction

Oops! Something went wrong

[next-mdx-remote-client] error compiling MDX: Expected a closing tag for `<tool_call>` (20:128-20:139) before the end of `paragraph` 18 | - Serving: supports vLLM for 1M-context serving and has GGUF quantizations for local runtimes; the YaRN factor creates a long-context vs short-context fidelity trade-off (factor 4.0 recommended only when you need the full 1M window). 19 | - Safety & usage: intentionally uncensored for research; deployers must provide review, access control, and policy filtering when exposing sensitive cybersecurity or biomedical capabilities. > 20 | - Format & tooling: ships a chat template compatible with Qwen3.5 function-call conventions so tool calls appear as structured <tool_call> blocks. | ^ 21 | More information: https://mdxjs.com/docs/troubleshooting-mdx

Back

Information

  • Websitehuggingface.co
  • OrganizationsEmpero AI, Alibaba (Qwen team)
  • Published date2026/07/13

Categories

Tags

More Items

Hugging Face
AI Model2026
Icon for item

WeMM-Embedding-9B

Junjie Zhou, Ke Mei +4WeChat Vision (Tencent Inc.), Tencent

Generates L2-normalized multimodal embeddings (default 4,096‑D) for text, images, videos and visual documents, supporting interleaved inputs and flexible dimension truncation (Matryoshka). Designed for cross-modal retrieval, ranking and downstream retrieval systems; audio is not supported.

multimodalembeddingsqwentransformerssafetensors+6
Hugging Face
AI Model2026
Icon for item

orcarouter/GLM-5.3-Flash-Uncensored-FP8

orcarouterorcarouter, zai-org (Z.ai / Zhipu AI)

Drop-in abliterated (refusal-removed) build of GLM-5.3-Flash that bakes refusal-direction removal into block-FP8 safetensors, yielding an uncensored 320B (18B active) multimodal MoE model with a 1M-token context. Intended for red-teaming, interpretability, and robustness research; MIT license; not for production without added guardrails.

moemultimodaltransformerssafetensorsfp8+6
Hugging Face
AI Model2026
Icon for item

DeepSeek-V4-Flash-Vision-Exp

DeepSeek AI

An experimental multimodal model that adds visual understanding to DeepSeek-V4-Flash: accepts text+image inputs and returns text analyses. Improves vision-dependent agent workflows while maintaining comparable text-only performance; released under an MIT license on Hugging Face.

deepseekmultimodalvisiontransformerssafetensors+7