WHAT’S NEW

New AI models.
Keep up without the noise.

The latest source listings from makers we track, covering language, images, video, audio, and 3D. Catalog checked Oct 29, 2026.

Sorted by the date a source repository was created, which may precede public release. This is a feed of new listings, not confirmed release announcements. Downloadable weights do not automatically mean an open-source license. Coverage and date meanings.

Language listings

1,615 dated records

Moonshot AI · Downloadable

Kimi-K2-Thinking

Text generationReasoning

Generates text for conversations and writing. Supports a reasoning mode.

Source listed
Nov 4, 2025
Parameters
1T
4-bit weights only
≈ 513.2 GB
License
other
Compare this model →

Microsoft · Downloadable

Fara-7B

Vision

Understands text and images and responds in text.

Source listed
Oct 30, 2025
Parameters
7B
4-bit weights only
≈ 3.5 GB
License
mit
Compare this model →

StepFun · Downloadable

Step-Audio-EditX

Text generation

Generates text, according to its published model architecture (Step1ForCausalLM). No task is listed on Hugging Face, so check the model card for intended use.

Source listed
Oct 29, 2025
Parameters
3.5B
4-bit weights only
≈ 1.8 GB
License
Not disclosed
Compare this model →

OpenBMB · Downloadable

InfLLM-V2-Short-Dense-Base

Text generation

Generates text, according to its published model architecture (MiniCPMForCausalLM). No task is listed on Hugging Face, so check the model card for intended use.

Source listed
Oct 25, 2025
Parameters
8.2B
4-bit weights only
≈ 4.1 GB
License
apache-2.0
Compare this model →

Z.ai · Downloadable

Glyph

Vision

Understands text and images and responds in text.

Source listed
Oct 25, 2025
Parameters
10.3B
4-bit weights only
≈ 5.1 GB
License
mit
Compare this model →

OpenBMB · Downloadable

InfLLM-V2-Long-Sparse-Base

Text generation

Generates text, according to its published model architecture (MiniCPMForCausalLM). No task is listed on Hugging Face, so check the model card for intended use.

Source listed
Oct 24, 2025
Parameters
8.2B
4-bit weights only
≈ 4.1 GB
License
apache-2.0
Compare this model →

MiniMax · Downloadable

MiniMax-M2

Text generation

Generates text for conversations and writing.

Source listed
Oct 22, 2025
Parameters
228.7B
4-bit weights only
≈ 114.3 GB
License
other
Compare this model →

Liquid AI · Downloadable

LFM2-VL-3B

Vision

Understands text and images and responds in text.

Source listed
Oct 22, 2025
Parameters
3B
4-bit weights only
≈ 1.5 GB
License
other
Compare this model →

Meta · Downloadable

MobileLLM-Pro-base

Text generation

Generates text for conversations and writing.

Source listed
Oct 21, 2025
Parameters
1.1B
4-bit weights only
≈ 0.5 GB
License
fair-noncommercial-research-license
Compare this model →

Datalab · Downloadable

chandra

VisionDocument reading

Understands text and images and responds in text.

Source listed
Oct 21, 2025
Parameters
8.8B
4-bit weights only
≈ 4.4 GB
License
openrail
Compare this model →

LightOn · Downloadable

LightOnOCR-1B-1025

VisionDocument reading

Understands text and images and responds in text.

Source listed
Oct 20, 2025
Parameters
1B
4-bit weights only
≈ 0.5 GB
License
apache-2.0
Compare this model →

Qwen · Alibaba · Downloadable

Qwen3-VL-2B-Thinking

VisionReasoning

Understands text and images and responds in text. Supports a reasoning mode.

Source listed
Oct 19, 2025
Parameters
2B
4-bit weights only
≈ 1 GB
License
apache-2.0
Compare this model →

What makes this feed useful?

Capability tags distinguish a model that reads images from one that creates them. Each listing links to its source and a permanent profile, so you can inspect its license and compare it with a model you already know.

Does this cover every new release?

No. The current feed tracks public Hugging Face metadata from selected makers. It can miss untracked makers, cloud-only releases, and records beyond a source’s retrieval limit. Announced models without a source listing are not included. Check Sources & methodology for the current coverage.

Listings update with the catalog automatically. A missing or delayed source can leave older information visible; each model profile shows when it was checked.