Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
Python agentaichatgptdeepseekfine-tuninggemmaimage-generationllamallmllmsopenaipythonqwenreinforcement-learningself-hostedstable-diffusiontext-to-speechttsuiunsloth
Commits on main
Studio: FLUX.3 prompt upsampling (Enhance prompt) through a llama.cpp sidecar (#14137)
def3451 Daniel Han committed 1 month ago
Studio: video families default to a hosted quantized text encoder (#14000)
9877243 Daniel Han committed 1 month ago
Studio: reserve MiniMax-H3's pinned weights from sd.cpp's budget on integrated AMD GPUs (#13786)
4232f63 oobabooga committed 1 month ago
Studio (macOS): size image loads against the Metal working set, not RAM minus 20% (#13776)
4b0e815 oobabooga committed 1 month ago
Studio: FLUX.3 Image Klein on the native sd.cpp engine (#14144)
2429c7a Daniel Han committed 1 month ago
Studio: keep Claude's web search to the sites it asks for (#14046)
20531d9 Nilay committed 1 month ago
Studio: install from a package index that lags PyPI (#14133)
888ccc5 Daniel Han committed 1 month ago
Studio: FLUX.3 Image Klein (distilled and base), with INT8 / FP8 / GGUF support (#13933)
5019141 Daniel Han committed 1 month ago
Load Ling 3.x low-rank queries on native bailing_hybrid and fix its exports (#14136)
32980a4 Daniel Han committed 1 month ago
Fix LM-head weight fake quantization in fused CE (#14006)
5f8a2ec cherry77-cloud committed 1 month ago
Read the current FSDP2 output head when scoring GRPO log-probs (#14060)
38e66cc cherry77-cloud committed 1 month ago
Studio: ask before ffmpeg or pandoc write outside the sandbox (#13942)
5e3b44d Nilay committed 1 month ago
Studio: wrap long commands and cap tall API examples in Settings (#14071)
4fe13f9 Etherl committed 1 month ago
Kernel follow-ups: NVFP4 shape-aware fused gate, MoE fallback tiles, test fixes across torch 2.10 - 2.14 (#14122)
4487b0a Daniel Han committed 1 month ago
Fix Nemotron loss normalization across gradient accumulation (#14002)
7ec86bb cherry77-cloud committed 1 month ago
Studio: ask before risky calls when a chat folder is too large to check (#13914)
6d203e3 Nilay committed 1 month ago
Studio: name a closed AMD device node in the System tab GPU notice (#14126)
ca3d1d0 Daniel Han committed 1 month ago
Studio: default to the Vulkan llama.cpp bundle when a CPU-bound host has a Vulkan GPU (#13599)
6a31011 Daniel Han committed 1 month ago
Studio: keep llama.cpp and vLLM connections' default system prompt (#14043)
638f134 Nilay committed 1 month ago
Studio: unload a local model without stopping external provider chats (#13917)
6f4a3e3 Nilay committed 1 month ago
Studio: keep GGUF variant listings off the default thread pool so a stalled Hub cannot block downloaded models (#14134)
336590e Daniel Han committed 1 month ago
Model picker contract: follow the On Device filter onto Discover's capability picks (#14138)
9c52c6f Daniel Han committed 1 month ago
Studio: let the Project sources panel be hidden in a project chat (#14125)
055f121 Daniel Han committed 1 month ago
Studio: skip a Terminal PATH entry MXC cannot grant instead of failing every launch (#14129)
c980574 Daniel Han committed 1 month ago
Load six older models in 4bit instead of a missing repo (#13941)
af0e2e1 Nilay committed 1 month ago
Unsloth Studio / Desktop (AMD, Windows): stop amd-smi from popping UAC/DiskPart prompts on the System page (#14041)
5ff4b87 Leo Borcherding committed 1 month ago