utLiquid AI · Downloadable
LFM2.5-Audio-1.5B-JP
Works with speech or audio. Check the model card for transcription and generation capabilities.
- Source listed
- May 26, 2026
- Parameters
- 1.5B
- 4-bit weights only
- ≈ 0.8 GB
- License
- other
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.
Latest listings
3,176 dated recordsLiquid AI · Downloadable
LFM2-2.6B-Longevity
Generates text for conversations and writing.
- Source listed
- May 25, 2026
- Parameters
- 2.6B
- 4-bit weights only
- ≈ 1.3 GB
- License
- other
Liquid AI · Downloadable
LFM2-1.2B-Longevity
Generates text, according to its published model architecture (Lfm2ForCausalLM). No task is listed on Hugging Face, so check the model card for intended use.
- Source listed
- May 25, 2026
- Parameters
- 1.2B
- 4-bit weights only
- ≈ 0.6 GB
- License
- other
OpenMOSS · Downloadable
MOSS-SoundEffect-v2.0
Works with speech or audio. Check the model card for transcription and generation capabilities.
- Source listed
- May 25, 2026
- Parameters
- 1.4B
- 4-bit weights only
- ≈ 0.7 GB
- License
- apache-2.0
OpenMOSS · Downloadable
MOSS-TTS-v1.5
Works with speech or audio. Check the model card for transcription and generation capabilities.
- Source listed
- May 25, 2026
- Parameters
- 8.5B
- 4-bit weights only
- ≈ 4.2 GB
- License
- apache-2.0
inclusionAI · Downloadable
SingGuard-4b
A specialized research model. See the model card for its intended task and supported setup.
- Source listed
- May 25, 2026
- Parameters
- 4B
- 4-bit weights only
- ≈ 2 GB
- License
- apache-2.0
inclusionAI · Downloadable
SingGuard-8b
A specialized research model. See the model card for its intended task and supported setup.
- Source listed
- May 25, 2026
- Parameters
- 8B
- 4-bit weights only
- ≈ 4 GB
- License
- apache-2.0
inclusionAI · Downloadable
SingGuard-2b
Understands text and images and responds in text.
- Source listed
- May 25, 2026
- Parameters
- 2B
- 4-bit weights only
- ≈ 1 GB
- License
- apache-2.0
NVIDIA · Downloadable
Cosmos-AnomalyGen-Glass-2B
A specialized research model. See the model card for its intended task and supported setup.
- Source listed
- May 25, 2026
- Parameters
- 2B
- 4-bit weights only
- ≈ 1 GB
- License
- other
NVIDIA · Downloadable
Cosmos-AnomalyGen-Metal-2B
A specialized research model. See the model card for its intended task and supported setup.
- Source listed
- May 25, 2026
- Parameters
- 2B
- 4-bit weights only
- ≈ 1 GB
- License
- other
NVIDIA · Downloadable
Cosmos-AnomalyGen-PCB-2B
Creates or edits images. Check the model card for supported inputs and output settings.
- Source listed
- May 25, 2026
- Parameters
- 2B
- 4-bit weights only
- ≈ 1 GB
- License
- other
NVIDIA · Downloadable
Nemotron-3-Super-120B-A12B-BF16-MTPv2
Generates text for conversations and writing.
- Source listed
- May 24, 2026
- Parameters
- 120B
- 4-bit weights only
- ≈ 60 GB
- License
- other
Google · Downloadable
gemma-4-12B
Hugging Face lists its task as any-to-any (several input and output types). See the model card for supported inputs and setup.
- Source listed
- May 23, 2026
- Parameters
- 12B
- 4-bit weights only
- ≈ 6 GB
- License
- apache-2.0
StepFun · Downloadable
Step-3.7-Flash
Understands text and images and responds in text.
- Source listed
- May 23, 2026
- Parameters
- 201.4B
- 4-bit weights only
- ≈ 100.7 GB
- License
- apache-2.0
Google · Downloadable
gemma-4-12B-it-assistant
Hugging Face lists its task as any-to-any (several input and output types). See the model card for supported inputs and setup.
- Source listed
- May 23, 2026
- Parameters
- 12B
- 4-bit weights only
- ≈ 6 GB
- License
- apache-2.0
Google · Downloadable
gemma-4-12B-it
Hugging Face lists its task as any-to-any (several input and output types). See the model card for supported inputs and setup.
- Source listed
- May 23, 2026
- Parameters
- 12B
- 4-bit weights only
- ≈ 6 GB
- License
- apache-2.0
NVIDIA · Downloadable
Nemotron-3.5-Content-Safety
A specialized research model. See the model card for its intended task and supported setup.
- Source listed
- May 22, 2026
- Parameters
- 4.3B
- 4-bit weights only
- ≈ 2.2 GB
- License
- other
NVIDIA · Downloadable
Cosmos3-Super-Image2Video
Generates video. Resolution, clip length, and speed depend on the version and setup.
- Source listed
- May 21, 2026
- Parameters
- 64.6B
- 4-bit weights only
- ≈ 32.3 GB
- License
- other
NVIDIA · Downloadable
Cosmos3-Nano-Policy-DROID
A specialized research model. See the model card for its intended task and supported setup.
- Source listed
- May 21, 2026
- Parameters
- 15.7B
- 4-bit weights only
- ≈ 7.9 GB
- License
- other
PrismML · Downloadable
bonsai-image-ternary-4B
Creates or edits images. Check the model card for supported inputs and output settings.
- Source listed
- May 21, 2026
- Parameters
- 4B
- 4-bit weights only
- ≈ 2 GB
- License
- apache-2.0
OpenBMB · Downloadable
MiniCPM5-1B
Generates text for conversations and writing.
- Source listed
- May 21, 2026
- Parameters
- 1B
- 4-bit weights only
- ≈ 0.5 GB
- License
- apache-2.0
OpenBMB · Downloadable
MiniCPM5-1B-SFT
Generates text for conversations and writing.
- Source listed
- May 21, 2026
- Parameters
- 1B
- 4-bit weights only
- ≈ 0.5 GB
- License
- apache-2.0
OpenBMB · Downloadable
MiniCPM5-1B-Base
Generates text for conversations and writing.
- Source listed
- May 21, 2026
- Parameters
- 1B
- 4-bit weights only
- ≈ 0.5 GB
- License
- apache-2.0
Qwen · Alibaba · Downloadable
Qwen-Image-Bench
Understands text and images and responds in text.
- Source listed
- May 21, 2026
- Parameters
- 27.4B
- 4-bit weights only
- ≈ 13.7 GB
- License
- apache-2.0
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.