THE MODEL DIRECTORY
Explore models. MY HARDWARE Your machine Choose your actual machine to personalize model and workflow ratings.
Choose my machine Windows / Linux PC A desktop or laptop GPU Mac Apple silicon and unified memory DGX Spark 1–4 linked NVIDIA systems Rented GPU A cloud machine you choose CPU only No dedicated GPU
Save my machine
Saved only in this browser. Model cards show a machine-specific rating when the setup supports one, or explain what evidence is still needed.
TASK → MACHINE → MODEL
Find your fit What do you want to do? Explore all tasks Qwen3.5 · 9B private chat setup Build a private coding assistant with Qwen3.8 27B Create images locally with FLUX.2 Klein Generate videos locally with Wan2.1 Review security with DeepSeek V4.1 Flash and Hermes Run a private reasoning assistant with Bonsai 2 · 27B Edit images locally with FLUX.2 Klein Create MiniMax H3 video on a rented GPU Ask your documents locally with Qwen3.5 · 9B Build a private coding assistant on rented GPUs Ask questions about an image privately with Gemma 4 Build a local AI tool agent with Qwen3 8B Extract structured data locally with gpt-oss 20B Create transparent images locally with Qwen-Image 2.1 Review code privately with Qwen3-Coder-Next on rented GPUs Transcribe an audio file privately Set up private chat with Bonsai 2 · 27B Build a local coding assistant with Bonsai 2 · 27B Build a local tool-using agent with Bonsai 2 · 27B Set up private Gemma 4 chat Generate a local video with LTX 2.5 Fast Build a local AI agent team with Qwen3 8B Generate MiniMax H3 video with audio in ComfyUI Animate a character with Wan Animate 2 Run RavenX CyberAgent for Authorized AI Pen Testing Run GLM-5.3 with Strix for Authorized Security Reviews Run GLM-5.3 UNCENSORED through an API Build a coding assistant with Strata (Qwen3.8-Flash-Next Coder) Generate MiniMax H3 videos with FreeVideo Create Uncensored AI Bots with Rakazo Create images locally with Qwen-Image-2.1 Edit images locally with Qwen-Image-2.1 Create video locally with HunyuanVideo 1.5 Build a private AI agent with GLM-5.3 Flash Code with GLM-5.3 Flash in VS Code + Cline Code locally with Mellum2.1 Thinking Build a private AI agent with Kolibri 1 Create a song locally with YuE2 3B Build a local AI agent with Qwen3.8 27B Build a private AI agent with Qwen3.8-Flash-Next Code with Qwen3.8-Flash-Next in VS Code + Cline Create short video with audio using Kandinsky 6 Pro Create a physical-AI scene with Cosmos3 Super Set up private phone chat with LFM3-1.7B Uncensored Set up private desktop chat with LFM3-1.7B Uncensored Run DeepSeek V4.1 Flash Uncensored with Strata Run a private reasoning assistant with DeepSeek R1 · 14B Set up private chat with Gemma 3 · 4B Set up private chat with Gemma 4 · 12B Set up private chat with DeepSeek R1 · 14B Set up private chat with gpt-oss · 120B Run a private reasoning assistant with gpt-oss · 120B Set up private chat with gpt-oss · 20B Run a private reasoning assistant with gpt-oss · 20B Set up private chat with Qwen2.5 Coder · 7B Chat locally with Qwen3.5 · 4B Set up private chat with Qwen3.8-27B Chat locally with MiniCPM5 · 2B Chat privately with Nemotron 3 Nano Omni Build a local AI agent with Ministral 3 Reasoning Build a local coding assistant with MiniCPM5 2B SFT Chat privately with uncensored Gemma 4 E4B Chat locally with Qwen3-Coder-Next openbmb/MiniCPM5-2B-GGUF MiniCPM5 2B local coding assistant metadata plan LFM2.5 2.6B local agent setup Build a coding assistant with GLM 5.2 on rented GPUs Build a local coding assistant with Qwen3-Coder-Next Build a local AI agent with Qwen3-Coder-Next Animate images locally with Wan2.1 Transcribe audio locally with Whisper large-v3-turbo Generate speech locally with Kokoro-82M Generate music locally with ACE-Step 1.5 Analyze images locally with Qwen3.5 Build private search with Qwen3 Embedding Improve local search with Qwen3 Reranker Create images on a rented GPU with SDXL Turbo Model variant Publisher package when known Q4 · lower memory Q8 · higher quality FP16 · original weights Context 4K tokens 8K tokens 32K tokens 64K tokens Experience All models Great only Runs well or better Runs well only Likely runs Slow Won’t run locally Setup source only
Choose a specific task to see its recommended model and documented alternatives. A memory-fit estimate is distinct from a verified run setup. Estimated ratings name their package and assumptions; source-only cards link to the available run path without inventing a hardware verdict.
All categories Text Coding Images Video Audio Agents
Model freshness Any model age Last 30 days Last 90 days Last 6 months Last year Model popularity Any popularity 40+ · gaining traction 60+ · popular 80+ · widely used Arena category Text Agents Web coding Vision Documents Image generation Image editing Text to video Image to video Video editing Arena · Open Source rank All models Ranked models only Top 5 Top 10 Top 25 Top 50 Not ranked in this category How scores work ↗ Freshness uses release date, or repository age when no release date is available. Rank filters use the selected Arena category, with #1 ranked highest. Unranked models remain visible unless you filter them out. Hardware and score filters must match the same version.
Publisher Any publisher ACE-Step Aleph Alpha Altworld AtomicChat · Liquid AI base model Black Forest Labs ByteDance-Seed CohereLabs DeepSeek Google Google DeepMind HauhauCS HuggingFaceTB IFM JetBrains Kandinsky Lab Lightricks LiquidAI MiniMax Mistral AI Moonshot AI Multimodal Art Projection NVIDIA OpenAI OpenVDN Prism ML Qwen Qwen base · Strata runtime · ISTA-DASLab Coder derivative RavenX Stability AI Tencent TokenRhythm Wan Wan-AI XHToken XiaomiMiMo XingChen-AGI Z.ai agentionai black-forest-labs byteshape dealignai dealignai · based on Z.ai GLM-5.3 deepseek-ai hexgrad huihui-ai ibm-granite jinaai krea microsoft mistralai nvidia openai openbmb openjev ornith-ai prism-ml stepfun-ai tencent ukisai yandex zai-org License Any license Apache 2.0 Apache-2.0 CC BY-NC 4.0 · noncommercial GLM-5.3 License Gemma Terms Kimi K3 License LFM Open License v1.0 · see the linked terms MIT MIT; base-model terms apply MiniMax H3 Community License Agreement OpenMDW 1.1 Qwen Community License 1.0 Qwen Community License 1.0 (base); MIT (Strata runtime); derivative terms apply Qwen Research License Stability AI Non-Commercial Community License; commercial use requires separate terms Tencent Hunyuan Community License Upstream GLM-5.3 License; derivative repository declares MIT apache-2.0 cc-by-nc-4.0 gemma krea-2-community-license lfm1.0 ltx-2.x-community-license-agreement minimax-h3-community-license-agreement mit nvidia-open-model-agreement openmdw-1.1 swift-open-license-1.0 Capability Any capability API ASR Cline ComfyUI Diffusers English FP8 GGUF GLM-5.3 German LFM3 MoE Ollama RunPod Strata TTS agentic agents animation any-to-any assistant audio automatic-speech-recognition character chat code code completion coding comfyui community derivative compressed weights cybersecurity diffusion distilled embeddings expert offload feature-extraction gguf image editing image generation image understanding image-generation image-text-to-text image-to-image image-to-text image-to-video instruct llama.cpp local audio local chat local video long context low VRAM lyrics math multilingual multimodal music generation music-generation offline phone AI physical AI rag realtime reasoning rented GPU reranking runtime route safetensors security review self-hosted semantic-search sentence-similarity speech recognition speech-to-text structured output structured prompts text-generation text-ranking text-to-image text-to-music text-to-speech text-to-video tool calling tools transcription transparency uncensored uncensored variants vLLM video vocals voice zero-shot-classification Deployment Local or rented GPU Own hardware Rented GPU Reset filters
8 models Sorted by release recency, documented run path, and source quality.
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. Freshness 69/100 96 days old Popularity 63 out of 100. 58,712 downloads in 30 days; 114 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality. Popularity 63/100 58.7K downloads / 30d
MODEL SIZE 0.47B
Mistral AI Voxtral Mini · 4B Realtime Voxtral Mini 4B Realtime 2602 is a multilingual, realtime automatic speech recognition model from Mistral AI, released under the Apache-2.0 license. It features a natively streaming architecture with a causal audio encoder and sliding window attention, enabling configurable transcription delays from 240ms to 2.4s (and reportedly 80ms to 2.4s). The model has approximately 4B parameters (≈3.4B language model and ≈970M audio encoder) and is described as optimized for on-device deployment. It supports 13 languages and is presented as achieving accuracy comparable to offline systems with less than 500ms delay, specifically matching leading offline open-source models and realtime APIs at a 480ms delay. The model card recommends a temperature of 0.0, using websockets for audio streaming, and a delay of 480ms as a balance between performance and latency.
Freshness 33 out of 100. Repository age: 291 days old, Jan 21, 2026. Measures age, not quality. Freshness 33/100 291 days old Popularity 83 out of 100. 1,473,664 downloads in 30 days; 1,060 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality. Popularity 83/100 1.5M downloads / 30d
MODEL SIZE 4B
openai Whisper Explore 6 versions of Whisper. Compare model sizes, packages, hardware requirements, and available setups.
6 versions Compare sizes & formats
Showing whisper base
Freshness 0 out of 100. Repository age: 1,504 days old, Sep 26, 2022. Measures age, not quality. Freshness 0/100 1,504 days old Popularity 86 out of 100. 4,305,917 downloads in 30 days; 313 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality. Popularity 86/100 4.3M downloads / 30d
MODEL SIZE 0.07B