openjutsu
One-click character swap for videos: an open-source, local Higgsfield Genjutsu alternative built on ComfyUI and MiniMax H3, with FLUX.3 Klein relighting.
- 2.6k
- TypeScript
- MIT
- 3 wk ago
One-click character swap for videos: an open-source, local Higgsfield Genjutsu alternative built on ComfyUI and MiniMax H3, with FLUX.3 Klein relighting.
Open decision models from 0.8B to 27B: a calibrated probability for every option in one pass, and a short thought only when unsure. Engine for vLLM, llama.cpp and MLX.
Open-source Higgsfield on ComfyUI: 120 image, video, VFX and audio apps, every model open weights; runs on Modal or your own GPU
The open creative studio on ComfyUI: images, video, 3D and audio on open weights, on your own GPU or the cloud. Camera motions, effects, apps, Cinema, Layers, Image to 3D and an AI agent.
Open-source Higgsfield AI alternatives on GitHub: 117 projects, 110 of them from ComfyUI's #OpenSourceHiggsfield challenge, with its 16 prize winners. Character swap (OPENjutsu), effects, camera re-shoots, films and full studios, with stars and licenses. CC0.
Every uncensored and abliterated Kimi build — Kimi K3.1 Pro Uncensored via API at 0.30/1.50 USD per million tokens, abliterated open-weight checkpoints for K3.1 Pro and K3, license notes and a red-team prompt library. Maintained by RouterHub.
Open-source Higgsfield on your own GPU: train a recurring character, turn a one-line brief into an edited reel, move the camera, re-shoot an angle, block shots in 3D, sketch to video, restore old footage. Runs on local ComfyUI, open-weights models only.
Open-source alternative to TypeSafe's Jev: a System One style model layer that gives typed, calibrated decisions from any open-weights LLM in one forward pass (HF + vLLM), with honest benchmarks
Real browser automation with local GLiNER2 inference and open weights. Work in progress. Contributions welcome.
Same take, new cast: open-weights video character recasting on local ComfyUI with SAM 3.1, FLUX 3 Image Klein and SCAIL-2. Built for #OpenSourceHiggsfield.
Every uncensored and abliterated DeepSeek build — V4.1 Flash 1017 Abliterated (the pinned, reproducible build that ranks third on the AA Cyber benchmark), V4 Pro, the tuned variants, abliterated GGUF builds and a cyber red-team prompt library. Maintained by RouterHub.
Hemmingway-1: a 27B model that writes the way a person writes. Open weights, free for non-commercial use (CC BY-NC 4.0); commercial use by agreement.
MODA: open fashion retrieval benchmark and models by Hopit AI. MODA (203M, open source), MODA Pro Lite (213M, open weights), MODA Pro (hosted). Full-corpus benchmarks vs FashionSigLIP, SigLIP-SO400M and ZooClaw — one harness, losses shown. #1 open model on LookBench.
SpEdE, the Splat Edit Engine: click an object in a Gaussian splat, then Remove, Replace or Extract it. Local ComfyUI nodes + web Studio, open weights only. #OpenSourceHiggsfield
Render it ten ways, keep the one you love. An open-weights film studio for MiniMax H3 in ComfyUI: takes, lineage, cast and an agent skill.
Curated Ideogram 4.0 prompts and image examples — typography, photorealistic portraits, product/UI mockups, and model comparisons. By Evolink.
A community-maintained catalog of uncensored and low-filter AI video models, including Seedance Spicy, Wan Spicy, and MiniMax H3 Spicy API variants, local weights, adapters, hosted endpoints, and practical trade-offs.
Turn a song and one line about its world into a lip-synced music video. ComfyUI nodes and a Studio app that run locally on open-weights models.
Introducing Muse Glimmer: open-weight 30B agentic multimodal model that runs on your device (Meta). Interactive local agent lab + guide. Apache 2.0 · on-device AI · function calling
OpenMayhem is an open-source and decentralized inference router for consumer & pro hardware. Providers earn, users/agents access cheap inference.
Read which GPU class initialised an open-weight LLM from its never-trained embedding rows. Independent replication and extension of Checkpoint Forensics.
A community-maintained catalog of open and low-filter AI image models, checkpoints, and local inference resources.
Self-hosted, Jev-compatible decision-model API — one /v1/systemone endpoint, open weights (Laya, kev), one Docker image per engine.
Static JevBench snapshot for AI decision models
Runs in the browser on Hugging Face
A local launcher and platform for open-weight generative models. Install video, image and speech studios from their own repositories, run them on your Mac, and keep everything they make in one gallery.
Did the agent actually finish? Inspectable GLiNER2.5-Decide fine-tuning on task completion and evidence spans.
An independent, evidence-graded read on how much you own the open models and inference providers you rely on.
Least-data SFT on Qwen2.5-0.5B-Instruct: teach a ticket-card format with 8 examples. Colab T4 notebook + LoRA. Open weights.
Laya 0.4B: The Free Jev Killer That Answers in 32ms! Sub-35ms open-weight System 1 decision model running real-time Pong AI.
Compare a fine-tune with its base: the weights list which vocabulary tokens its training data contained. 421 release/base comparisons across 580 open models.
Independent field guide to Salesforce SOV-Transcribe-1: the launch table, the prompts, and what it takes to run the weights.
Official landing page for Agnes-Video-2.0-MoE audio-video generation weights and inference resources.
Abliterated & uncensored LLM coding benchmarks: 74 open-weight models/quants (Kimi K2.7, K3, GLM-5.3, MiMo, DeepSeek-V4.1, Qwen3.8, Qwen3-Coder) on SWE-bench Verified, refusal rate, tok/s on Mac Studio M3 Ultra + RTX PRO 6000, vs Claude/Codex
Measures whether an LLM's self-reports agree with what its internals show. Injected concepts are perfectly recoverable by probes (AUC 1.000) and absent from the model's verbal report - report-state agreement fails in a graded way. Gemma-3-4B, Jacobian lens, ~3,750 trials.