Get the latest setups
THE PRACTICAL AI FIELD GUIDE

Find the right AI.
Run it your way.

Models, workflows, hardware requirements, and cloud GPU options. Everything between “heard about it” and “running it.”

Get the latest setups Explore models
Practical setups. Open choices. No hype.
YOUR AI. YOUR INFRASTRUCTURE.
01 / FIND YOUR MODEL

Mellum2.1 Thinking

12B total · 2.5B active parameters · Open weights

Explore local setup

Check requirements and official guidance.

Run in the cloud

More power. On demand.

02 / CHOOSE HOW YOU RUNNO LOCK-IN ↗
BUILT FOR PEOPLE WHO BUILDLOCAL + CLOUDOPEN WEIGHTSREAL-WORLD WORKFLOWS

What do you want to do?

Explore the directory

Text

Chat, write, and reason

Coding

Code on your terms

Images

Create and edit

Video

Turn ideas into motion

Audio

Transcribe and speak

Agents

Connect and automate

Compute

Find your GPU

RECENTLY ADDED

Current models to explore.

Downloadable weights, documented run paths, and clear evidence for hardware fit.

All models
Text
XiaomiMiMo

MiMo V2.6

Explore 3 versions of MiMo V2.6. Compare model sizes, packages, hardware requirements, and available setups.

3 versionsCompare sizes & formats
ShowingMiMo V2.6 Distill Qwen 9B
Freshness 83 out of 100. Repository age: 48 days old, Sep 21, 2026. Measures age, not quality.
Popularity 60 out of 100. 17,769 downloads in 30 days; 694 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
MODEL SIZE9.41B
LOCALRENTED GPU
Compare versions →
Coding
XingChen-AGI

Xing4.0

Explore 3 versions of Xing4.0. Compare model sizes, packages, hardware requirements, and available setups.

3 versionsCompare sizes & formats
ShowingXing4.0 29B A4B
Freshness 82 out of 100. Repository age: 53 days old, Sep 16, 2026. Measures age, not quality.
Popularity 56 out of 100. 5,343 downloads in 30 days; 1,865 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
MODEL SIZE31.22B
LOCALRENTED GPU
Compare versions →
Images
Qwen

Qwen-Image 2.1

Generate and edit images, including transparent images and edits guided by reference images.

Freshness 81 out of 100. Repository age: 55 days old, Sep 14, 2026. Measures age, not quality.
Popularity 76 out of 100. 248,964 downloads in 30 days; 3,575 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
Arena Open Source Image generation: rank 1 of 19. Ranked model: Qwen-Image-2.1. Export dated Nov 6, 2026. Source checked Nov 7, 2026. Human preference ranking, not hardware suitability.+1
MODEL SIZE7B visual generator
LOCALRENTED GPU
Repository opened Sep 14, 2026
Video
OpenVDN

vdn minimax h3

VDN-Minimax-H3 (VDN-H3) is a hybrid-attention text-to-video model built on MiniMax H3 AI/MiniMax-H3. Its repository provides safetensors weights for a diffusers-based text-to-video pipeline. The model card describes a fast-inference design using a hybrid-attention architecture: one frame-wise linear attention branch plus a softmax branch, along with two small LoRA adapters and a separate linear attention branch that can be merged into the backbone during inference without modifying backbone weights. Reported setup uses SGLang Diffusion on 8xB200 GPUs with 8 denoising steps. The license is the minimax-h3-community-license-agreement, and users are directed to read the license before downloading or running the model. The model card also links to a paper, blog, code repository, weights, and ModelScope page, and notes support for I2VA, FL2VA, and L2VA with the same checkpoint.

Freshness 77 out of 100. Repository age: 67 days old, Sep 2, 2026. Measures age, not quality.
Popularity 50 out of 100. 3,081 downloads in 30 days; 360 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
LOCALRENTED GPU
Repository opened Sep 2, 2026
Audio
ibm-granite

granite speech 5.0 470m turboctc

Granite Speech 5.0 470M TurboCTC is a compact 470 million parameter English automatic speech recognition model released by IBM under the Apache 2.0 license. It uses a conformer acoustic encoder with block self-attention, self-conditioning, and temporal downsampling, with an output layer of 16,384 BPE units. The model was trained on approximately 60,000 hours of English audio from public corpora using Connectionist Temporal Classification (CTC), and inference is non-autoregressive with greedy decoding. It is designed for low-latency and high-throughput English speech-to-text transcription, and is described as suitable for deployment on laptops, smartphones, and other edge devices. Native support is available in transformers>=5.16.0, with additional usage options for Apple Silicon via mlx-audio and local GGUF inference via transcribe.cpp.

Freshness 69 out of 100. Repository age: 96 days old, Aug 4, 2026. Measures age, not quality.
Popularity 63 out of 100. 58,712 downloads in 30 days; 114 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
MODEL SIZE0.47B
LOCALRENTED GPU
Repository opened Aug 4, 2026
Text
openjev

openjev

OpenJev is an open-weights decision model for typed zero-shot classification over text, web pages, and screenshots. It accepts labels at request time and produces choices, yes/no probabilities, or scores without free-form generation. The model supports multiple languages including English, German, French, Hindi, Chinese, and Japanese. Reported latency is about 80 ms for a short text decision and about 210 ms for an isolated web decision on one H100 under specified conditions.

Freshness 83 out of 100. Repository age: 49 days old, Sep 20, 2026. Measures age, not quality.
Popularity 55 out of 100. 9,770 downloads in 30 days; 188 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
MODEL SIZE27.36B
LOCALRENTED GPU
Repository opened Sep 20, 2026

Want these rated for your machine? Add My Hardware →

LESS NOISE. MORE KNOW-HOW.

AI moves too fast
for generic lists.

Useful information goes beyond a model name. Our setup records show what works, what breaks, and what it takes to run a model.

How we approach testing
01 /

The exact setup

Hardware, software, quantization, and context. Details you can reproduce.

02 /

Local or cloud. Your call.

Understand the tradeoffs before you spend on hardware or compute.

03 /

Evidence over claims

No invented scores. Tested, verified, and estimated claims are clearly distinguished.

TWO WAYS TO GET THERE

Your model. Your choice.

Keep it close.

Run AI on hardware you own.

  • Keep files on your machine
  • Use your own GPU
  • No hourly compute fees
  • Built for frequent use and privacy
Explore local setups

Go beyond your GPU.

Rent the hardware when you need it.

  • No expensive GPU required
  • Access more powerful hardware
  • Check compute and storage billing
  • An option for larger workloads
Explore cloud GPUs

Need this running inside your company?

Private models, internal knowledge, and workflows on infrastructure you control.

Explore teams