§01·tested · /gpus/rtx-3090
Recipes tested on RTX 3090
45 community-tested setups — recipes whose author ran them on this exact card.
nvidia30 series24GB VRAM
- multimodalintermediate24GB+recipe
Qwen3.8-27B on RTX 3090: 128K-context vision chat via llama.cpp GGUF
- multimodalintermediate24GB+recipe
Muse Glimmer 30B on RTX 3090: vision + DFlash speculation at full 131K context
- videoadvanced24GB+recipe
LTX-2.5 on RTX 3090: 22B audio-video on Ampere via the official ComfyUI INT8 checkpoint
- llmintermediate24GB+recipe
Nanbeige4.2-3B on RTX 3090: the full 256K context in VRAM with llama.cpp
- videoadvanced12GB+recipe
MiniMax H3 on RTX 3090: 5-second audio+video in ComfyUI on pruned int8
- multimodaladvanced24GB+recipe
Fara1.5-9B on RTX 3090: BF16 Computer-Use Agent with Vision under llama.cpp
- multimodaladvanced16GB+recipe
Fara1.5-4B on RTX 3090: BF16 Computer-Use Agent at a 131K Context with llama.cpp
- multimodaladvanced24GB+recipe
Fara1.5-27B on RTX 3090: local browser computer-use agent with llama.cpp
- multimodaladvanced24GB+recipe
Agents-A1 35B-A3B on RTX 3090: 128K Agentic Serving via llama.cpp, and What Vision Costs on 24 GB
- llmadvanced24GB+recipe
KAT-Coder V2.5 Dev on RTX 3090: 35B-A3B Agentic Coding at 128K Context via llama.cpp
- llmintermediate24GB+recipe
Gemma 4 12B on RTX 3090: Local Private Assistant via llama.cpp / Ollama (24GB)
- llmintermediate24GB+recipe
Phi-4 (14B) on RTX 3090: Local Private Assistant via llama.cpp / Ollama (24GB)
- llmintermediate24GB+recipe
Mistral Nemo 12B on RTX 3090: Local Private Assistant via llama.cpp / Ollama (24GB)
- llmintermediate24GB+recipe
Mistral Small 3.2 24B on RTX 3090: Local Private Assistant via llama.cpp / Ollama (24GB)
- llmintermediate24GB+recipe
Devstral Small 2 (24B) on RTX 3090: Local Agentic Coding via llama.cpp + OpenHands (24GB Value Tier)
- llmadvanced24GB+recipe
Laguna XS 2.1 on RTX 3090: Local Agentic Coding via Ollama / llama.cpp + OpenHands (24GB Value-Entry Tier)
- llmadvanced24GB+recipe
North Mini Code 1.0 on RTX 3090: Local Agentic Coding via llama.cpp + OpenHands (24GB Entry Tier)
- llmadvanced24GB+recipe
Ornith 1.0 35B on RTX 3090: Local Agentic Coding via llama.cpp + OpenHands (24GB Entry Tier)
- llmintermediate24GB+recipe
Qwen3-30B-A3B on RTX 3090: Full-GPU MoE Chat at 153 tok/s
- multimodalintermediate24GB+recipe
Qwen3.5-35B-A3B on RTX 3090: MXFP4 MoE Chat at 111 tok/s
- multimodalintermediate24GB+recipe
Qwen3.5-27B on RTX 3090: Q4_K GGUF local chat via llama.cpp
- llmintermediate24GB+recipe
Gemma 4 31B on RTX 3090: dense 31B local chat via Q4_K_M GGUF in Ollama / llama.cpp
- imageintermediate16GB+recipe
Krea 2 Turbo (FP8) on RTX 3090 via ComfyUI: 8-Step Text-to-Image, with the Raw Tier in Reach
- multimodalbeginner20GB+recipe
Gemma 4 E4B on RTX 3090: Multimodal Inference via BF16 (with 16 GB of Headroom to Spend)
- 3dintermediate12GB+recipe
Waypoint 1.5 on RTX 3090: Real-Time Interactive World Model at 720p, 30 FPS
- imagebeginner13GB+recipe
Flux.2 Klein 4B on RTX 3090: BFL-Recommended ~13 GB CPU-Offload Path for 4-Step Text-to-Image
- imageintermediate13GB+recipe
Qwen-Image on RTX 3090: 20B Text-to-Image via ComfyUI GGUF (Ampere Path — No FP8)
- videointermediate24GB+recipe
HunyuanVideo-1.5 on RTX 3090: 480p Step-Distilled Image-to-Video on a Razor-Thin 24 GB Envelope
- videointermediate24GB+recipe
Wan 2.2 T2V-A14B on RTX 3090: 720p text-to-video in ComfyUI with FP8 weights (Ampere)
- videointermediate14GB+recipe
LightX2V on RTX 3090: 4-Step Text-to-Video with Distilled Wan2.1-14B via INT8 / BF16 Offload
- videointermediate22GB+recipe
Mochi 1 on RTX 3090: 49-frame 480p Text-to-Video with Diffusers
- videointermediate14GB+recipe
CogVideoX 1.5 5B on RTX 3090: 1360x768 Text-to-Video with Diffusers
- videointermediate24GB+recipe
Wan 2.2 TI2V-5B on RTX 3090: 720p Text/Image-to-Video in ComfyUI
- llmintermediate22GB+recipe
Qwen3-32B on RTX 3090: UD-Q4_K_XL GGUF via llama.cpp
- llmintermediate18GB+recipe
Gemma 4 26B A4B-it on RTX 3090: Local Multimodal Chat via Q4_K_M GGUF + llama.cpp
- llmbeginner10GB+recipe
Qwen3-14B on RTX 3090: Q4_K_M GGUF via Ollama or llama.cpp
- llmbeginner12GB+recipe
DeepSeek-R1-Distill-Qwen-14B on RTX 3090 via Ollama Q4_K_M GGUF
- llmbeginner6GB+recipe
Qwen3-8B on RTX 3090: Q4_K_M GGUF with 18 GB of Headroom for Colocation or Long Context
- llmbeginner10GB+recipe
Llama 3.1 8B on RTX 3090: Local Chat via llama.cpp + Unsloth UD-Q4_K_XL GGUF
- llmbeginner16GB+recipe
gpt-oss 20B on RTX 3090: MXFP4 Chat at 147 tok/s via Ollama or vLLM
- imageintermediate20GB+recipe
HiDream-O1-Image Full BF16 on RTX 3090: 2048×2048 Text-to-Image in ComfyUI
- imageintermediate18GB+recipe
LongCat-Image (base T2I) on RTX 3090: Bilingual 6B Text-to-Image via diffusers BF16 + CPU Offload
- imageintermediate24GB+recipe
Chroma1-Base (V48) on RTX 3090: Uncensored 8.9B FLUX.1-Schnell De-Distillation via Diffusers BF16
- imageintermediate16GB+recipe
Juggernaut Z on RTX 3090: Cinematic Photoreal Fine-Tune of Z-Image Base at BF16 via Diffusers or ComfyUI
- imagebeginner16GB+recipe
Z-Image Turbo on RTX 3090: 8-Step 1024x1024 Text-to-Image at BF16 with Diffusers or ComfyUI