Datasets:

Smeltcore
/

gpu-compatibility

gpu_slug string	model_slug string	model_name string	vertical string	model_status string	fit string	min_vram_gb float64	peak_vram_gb float64	task string	speed_value float64	speed_unit string	works bool	confidence float64	recipe_count int64	recipe_slug string	url string
m2-max	acestep-1-5-xl	ACE-Step 1.5 XL	music	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/acestep-1-5-xl/
m2-max	agents-a1-35b-a3b	Agents-A1 35B-A3B	multimodal	active	unknown	24	null	null	null	null	null	null	0	null	https://smeltcore.com/models/agents-a1-35b-a3b/
m2-max	agents-a1-4b	Agents-A1 4B	multimodal	active	verified	14	null	null	null	null	true	null	1	agents-a1-4b-on-apple-m2-max-first-party-gguf-vision-on-metal-at-the-full-262k-context	https://smeltcore.com/recipes/agents-a1-4b-on-apple-m2-max-first-party-gguf-vision-on-metal-at-the-full-262k-context/
m2-max	anima	Anima	image	active	unknown	6	null	null	null	null	null	null	0	null	https://smeltcore.com/models/anima/
m2-max	animatediff	AnimateDiff	video	active	unknown	6	null	null	null	null	null	null	0	null	https://smeltcore.com/models/animatediff/
m2-max	apodex-1-1-mini	Apodex 1.1 mini	llm	active	verified	36	null	null	null	null	true	null	1	apodex-1-1-mini-on-apple-m2-max-6-bit-mlx-at-131k-and-why-262k-is-a-different-tier	https://smeltcore.com/recipes/apodex-1-1-mini-on-apple-m2-max-6-bit-mlx-at-131k-and-why-262k-is-a-different-tier/
m2-max	bonsai-27b	Bonsai 27B	multimodal	active	fits	6	null	null	null	null	null	null	0	null	https://smeltcore.com/models/bonsai-27b/
m2-max	chroma-v48	Chroma V48	image	active	unknown	10	null	null	null	null	null	null	0	null	https://smeltcore.com/models/chroma-v48/
m2-max	cogvideox-1-5	CogVideoX 1.5	video	active	unknown	10	null	null	null	null	null	null	0	null	https://smeltcore.com/models/cogvideox-1-5/
m2-max	deepseek-r1-distill-qwen-14b	DeepSeek R1 Distill 14B	llm	active	unknown	10	null	null	null	null	null	null	0	null	https://smeltcore.com/models/deepseek-r1-distill-qwen-14b/
m2-max	deepseek-v3	DeepSeek V3	llm	active	unknown	null	null	null	null	null	null	null	0	null	https://smeltcore.com/models/deepseek-v3/
m2-max	devstral-small-24b	Devstral Small 2 (24B)	llm	active	verified	48	null	null	null	null	true	null	1	devstral-small-2-24b-on-apple-m2-max-local-agentic-coding-via-llama-cpp-metal-openhands-64gb-apple-q	https://smeltcore.com/recipes/devstral-small-2-24b-on-apple-m2-max-local-agentic-coding-via-llama-cpp-metal-openhands-64gb-apple-q/
m2-max	ernie-image-turbo	ERNIE-Image-Turbo	image	active	unknown	9	null	null	null	null	null	null	0	null	https://smeltcore.com/models/ernie-image-turbo/
m2-max	falcon2-11b	falcon2 11b	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/falcon2-11b/
m2-max	fara1-5-27b	Fara1.5-27B	multimodal	active	verified	48	null	null	null	null	true	null	1	fara1-5-27b-on-apple-m2-max-browser-computer-use-agent-at-the-full-262k-context	https://smeltcore.com/recipes/fara1-5-27b-on-apple-m2-max-browser-computer-use-agent-at-the-full-262k-context/
m2-max	fara1-5-4b	Fara1.5-4B	multimodal	active	verified	16	null	null	null	null	true	null	1	fara1-5-4b-on-apple-m2-max-browser-computer-use-agent-on-llama-cpp-metal	https://smeltcore.com/recipes/fara1-5-4b-on-apple-m2-max-browser-computer-use-agent-on-llama-cpp-metal/
m2-max	fara1-5-9b	Fara1.5-9B	multimodal	active	verified	16	null	null	null	null	true	null	1	fara1-5-9b-on-apple-m2-max-browser-computer-use-agent-on-llama-cpp-metal	https://smeltcore.com/recipes/fara1-5-9b-on-apple-m2-max-browser-computer-use-agent-on-llama-cpp-metal/
m2-max	flux-1-dev	Flux.1 Dev	image	active	unknown	16	null	null	null	null	null	null	0	null	https://smeltcore.com/models/flux-1-dev/
m2-max	flux-2-klein-4b	Flux.2-Klein-4B	image	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/flux-2-klein-4b/
m2-max	foundation-1	Foundation-1	music	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/foundation-1/
m2-max	gemma-2-9b	Gemma 2 9B	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/gemma-2-9b/
m2-max	gemma-4-12b	Gemma 4 12B	llm	active	fits	16	null	null	null	null	null	null	0	null	https://smeltcore.com/models/gemma-4-12b/
m2-max	gemma-4-e4b	Gemma 4 E4B-IT	multimodal	active	verified	5	null	null	null	null	true	null	1	gemma-4-e4b-m2-max	https://smeltcore.com/recipes/gemma-4-e4b-m2-max/
m2-max	gemma-7b	gemma 7b	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/gemma-7b/
m2-max	gemma4-26b	Gemma 4 26B MoE	multimodal	active	unknown	10.92	null	null	null	null	null	null	0	null	https://smeltcore.com/models/gemma4-26b/
m2-max	gemma4-31b	Gemma4 31B	multimodal	active	unknown	20	null	null	null	null	null	null	0	null	https://smeltcore.com/models/gemma4-31b/
m2-max	gpt-oss-120b	gpt-oss 120B	llm	active	unknown	null	null	null	null	null	null	null	0	null	https://smeltcore.com/models/gpt-oss-120b/
m2-max	gpt-oss-20b	gpt-oss 20B	llm	active	verified	13	null	null	null	null	true	null	1	gpt-oss-20b-m2-max	https://smeltcore.com/recipes/gpt-oss-20b-m2-max/
m2-max	hidream-o1-image	HiDream-O1-Image	image	active	unknown	10	null	null	null	null	null	null	0	null	https://smeltcore.com/models/hidream-o1-image/
m2-max	hunyuan-3d	Hunyuan3D	3d	active	unknown	10	null	null	null	null	null	null	0	null	https://smeltcore.com/models/hunyuan-3d/
m2-max	hunyuan-video	HunyuanVideo 1.5	video	active	unknown	14	null	null	null	null	null	null	0	null	https://smeltcore.com/models/hunyuan-video/
m2-max	juggernaut-z	Juggernaut Z	image	active	unknown	12	null	null	null	null	null	null	0	null	https://smeltcore.com/models/juggernaut-z/
m2-max	kat-coder-v2-5-dev	KAT-Coder V2.5 Dev	llm	active	verified	48	null	null	null	null	true	null	1	kat-coder-v2-5-dev-on-apple-m2-max-35b-agentic-coding-via-llama-cpp-metal-64gb-unified-memory	https://smeltcore.com/recipes/kat-coder-v2-5-dev-on-apple-m2-max-35b-agentic-coding-via-llama-cpp-metal-64gb-unified-memory/
m2-max	kimodo	KiMoDo	specialized	active	unknown	3	null	null	null	null	null	null	0	null	https://smeltcore.com/models/kimodo/
m2-max	kokoro-tts	Kokoro TTS	tts	active	verified	1	null	null	null	null	true	null	1	kokoro-tts-m2-max	https://smeltcore.com/recipes/kokoro-tts-m2-max/
m2-max	krea-2	Krea 2	image	active	verified	24	null	null	null	null	true	null	1	krea-2-m2-max	https://smeltcore.com/recipes/krea-2-m2-max/
m2-max	laguna-xs-2-1	Laguna XS 2.1	llm	active	verified	24	null	null	null	null	true	null	1	laguna-xs-2-1-on-apple-m2-max-local-agentic-coding-via-ollama-openhands-64gb-apple	https://smeltcore.com/recipes/laguna-xs-2-1-on-apple-m2-max-local-agentic-coding-via-ollama-openhands-64gb-apple/
m2-max	lightx2v	LightX2V	video	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/lightx2v/
m2-max	llama-3-1-70b	Llama 3.1 70B	llm	active	unknown	32	null	null	null	null	null	null	0	null	https://smeltcore.com/models/llama-3-1-70b/
m2-max	llama-3-1-8b	Llama 3.1 8B	llm	active	verified	5	null	null	null	null	true	null	1	llama-3-1-8b-m2-max	https://smeltcore.com/recipes/llama-3-1-8b-m2-max/
m2-max	llama-3-2-1b	Llama 3.2 1B	llm	active	unknown	2	null	null	null	null	null	null	0	null	https://smeltcore.com/models/llama-3-2-1b/
m2-max	llama-3-3-70b	Llama 3.3 70B	llm	active	verified	40	null	null	null	null	true	null	1	llama-3-3-70b-m2-max	https://smeltcore.com/recipes/llama-3-3-70b-m2-max/
m2-max	llama2-13b	llama2 13b	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/llama2-13b/
m2-max	llama2-7b	llama2 7b	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/llama2-7b/
m2-max	llava-7b	llava 7b	multimodal	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/llava-7b/
m2-max	longcat-image	LongCat Image	image	active	unknown	13	null	null	null	null	null	null	0	null	https://smeltcore.com/models/longcat-image/
m2-max	ltx-2	LTX-2	video	active	unknown	24	null	null	null	null	null	null	0	null	https://smeltcore.com/models/ltx-2/
m2-max	ltx-2-5	LTX-2.5	video	active	unknown	15.54	null	null	null	null	null	null	0	null	https://smeltcore.com/models/ltx-2-5/
m2-max	ltx-video-2-3	LTX-2.3	video	active	verified	14	null	null	null	null	true	null	1	ltx-video-2-3-m2-max	https://smeltcore.com/recipes/ltx-video-2-3-m2-max/
m2-max	ltx-video-2b	LTX-Video 2B	video	active	unknown	10	null	null	null	null	null	null	0	null	https://smeltcore.com/models/ltx-video-2b/
m2-max	minimax-h3	MiniMax H3 (Hailuo 3)	video	active	verified	64	null	null	null	null	true	null	1	minimax-h3-on-apple-m2-max-native-mlx-video-with-synchronised-audio	https://smeltcore.com/recipes/minimax-h3-on-apple-m2-max-native-mlx-video-with-synchronised-audio/
m2-max	minimind-o	MiniMind-O	multimodal	active	unknown	4	null	null	null	null	null	null	0	null	https://smeltcore.com/models/minimind-o/
m2-max	mistral-nemo-12b	Mistral Nemo 12B	llm	active	fits	16	null	null	null	null	null	null	0	null	https://smeltcore.com/models/mistral-nemo-12b/
m2-max	mistral-small-3-2-24b	Mistral Small 3.2 24B	llm	active	verified	48	null	null	null	null	true	null	1	mistral-small-3-2-24b-on-m2-max-64gb-local-private-assistant-via-llama-cpp-ollama-on-apple-metal	https://smeltcore.com/recipes/mistral-small-3-2-24b-on-m2-max-64gb-local-private-assistant-via-llama-cpp-ollama-on-apple-metal/
m2-max	mochi-1	Mochi 1	video	active	unknown	20	null	null	null	null	null	null	0	null	https://smeltcore.com/models/mochi-1/
m2-max	moss-audio	MOSS-Audio	multimodal	active	unknown	11	null	null	null	null	null	null	0	null	https://smeltcore.com/models/moss-audio/
m2-max	muse-glimmer-30b	Muse Glimmer 30B	multimodal	active	verified	36	null	null	null	null	true	null	1	muse-glimmer-30b-on-apple-m2-max-8-bit-mlx-with-vision-and-the-dflash-drafter	https://smeltcore.com/recipes/muse-glimmer-30b-on-apple-m2-max-8-bit-mlx-with-vision-and-the-dflash-drafter/
m2-max	nanbeige4-2-3b	Nanbeige4.2 3B	llm	active	verified	16	null	null	null	null	true	null	1	nanbeige4-2-3b-on-apple-m2-max-128k-context-agentic-llm-via-llama-cpp-metal	https://smeltcore.com/recipes/nanbeige4-2-3b-on-apple-m2-max-128k-context-agentic-llm-via-llama-cpp-metal/
m2-max	nemotron-3-5-lightning-30b-a3b	NVIDIA Nemotron 3.5 Lightning 30B-A3B	llm	active	unknown	null	null	null	null	null	null	null	0	null	https://smeltcore.com/models/nemotron-3-5-lightning-30b-a3b/
m2-max	north-mini-code-1-0	North Mini Code 1.0	llm	active	verified	48	null	null	null	null	true	null	1	north-mini-code-1-0-on-apple-m2-max-local-agentic-coding-via-llama-cpp-metal-openhands-64gb-unified-	https://smeltcore.com/recipes/north-mini-code-1-0-on-apple-m2-max-local-agentic-coding-via-llama-cpp-metal-openhands-64gb-unified-/
m2-max	omnivoice	OmniVoice	tts	active	unknown	4	null	null	null	null	null	null	0	null	https://smeltcore.com/models/omnivoice/
m2-max	openaudio-s1-mini	OpenAudio S1 Mini	tts	active	unknown	5	null	null	null	null	null	null	0	null	https://smeltcore.com/models/openaudio-s1-mini/
m2-max	ornith-1-0-35b	Ornith 1.0 35B	llm	active	verified	48	null	null	null	null	true	null	1	ornith-1-0-35b-on-apple-m2-max-local-agentic-coding-via-llama-cpp-metal-openhands-64gb-unified-memor	https://smeltcore.com/recipes/ornith-1-0-35b-on-apple-m2-max-local-agentic-coding-via-llama-cpp-metal-openhands-64gb-unified-memor/
m2-max	ornith-1-0-9b	Ornith 1.0 9B	llm	active	fits	12	null	null	null	null	null	null	0	null	https://smeltcore.com/models/ornith-1-0-9b/
m2-max	phi-4	Phi-4	llm	active	fits	16	null	null	null	null	null	null	0	null	https://smeltcore.com/models/phi-4/
m2-max	qwen-image	Qwen-Image	image	active	verified	23	null	null	null	null	true	null	1	qwen-image-m2-max	https://smeltcore.com/recipes/qwen-image-m2-max/
m2-max	qwen2-5-14b	Qwen2.5 14B	llm	active	unknown	9	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen2-5-14b/
m2-max	qwen2-5-7b	Qwen2.5 7B	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen2-5-7b/
m2-max	qwen2-7b	qwen2 7b	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen2-7b/
m2-max	qwen3-14b	Qwen3 14B	llm	active	fits	9	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen3-14b/
m2-max	qwen3-30b-a3b	Qwen3 30B-A3B	llm	active	unknown	12	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen3-30b-a3b/
m2-max	qwen3-32b	Qwen3 32B	llm	active	verified	19	null	null	null	null	true	null	1	qwen3-32b-m2-max	https://smeltcore.com/recipes/qwen3-32b-m2-max/
m2-max	qwen3-4b	Qwen3-4B	llm	active	unknown	4	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen3-4b/
m2-max	qwen3-5-27b	Qwen3.5 27B	multimodal	active	unknown	20	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen3-5-27b/
m2-max	qwen3-5-35b	Qwen3.5 35B	multimodal	active	unknown	24	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen3-5-35b/
m2-max	qwen3-6-35b-a3b-mtp-ud-q4-k-xl-gguf	Qwen3.6 35B-A3B	multimodal	active	unknown	9.8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen3-6-35b-a3b-mtp-ud-q4-k-xl-gguf/
m2-max	qwen3-8-27b	Qwen3.8 27B	multimodal	active	verified	48	null	null	null	null	true	null	1	qwen3-8-27b-on-apple-m2-max-8-bit-mlx-vision-language-with-mtp-speculative-decoding	https://smeltcore.com/recipes/qwen3-8-27b-on-apple-m2-max-8-bit-mlx-vision-language-with-mtp-speculative-decoding/
m2-max	qwen3-8b	Qwen3-8B	llm	active	fits	5	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen3-8b/
m2-max	qwen3-next-80b-a3b	Qwen3-Next 80B-A3B	llm	active	verified	48	null	null	null	null	true	null	1	qwen3-next-80b-a3b-on-apple-m2-max-an-80b-moe-assistant-in-64gb-unified-memory	https://smeltcore.com/recipes/qwen3-next-80b-a3b-on-apple-m2-max-an-80b-moe-assistant-in-64gb-unified-memory/
m2-max	qwen3tts	Qwen3-TTS	tts	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/qwen3tts/
m2-max	sam-3	SAM 3	specialized	active	unknown	4	null	null	null	null	null	null	0	null	https://smeltcore.com/models/sam-3/
m2-max	sd1-5	SD1.5	image	active	unknown	4	null	null	null	null	null	null	0	null	https://smeltcore.com/models/sd1-5/
m2-max	sdxl	Stable Diffusion XL	image	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/sdxl/
m2-max	sensenova-u1	SenseNova U1	image	active	unknown	12	null	null	null	null	null	null	0	null	https://smeltcore.com/models/sensenova-u1/
m2-max	stablelm2-12b	stablelm2 12b	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/stablelm2-12b/
m2-max	star-elastic-12b-nvfp4	Nemotron Elastic 12B (NVFP4)	llm	active	unknown	null	null	null	null	null	null	null	0	null	https://smeltcore.com/models/star-elastic-12b-nvfp4/
m2-max	sulphur-2	Sulphur 2	video	active	unknown	16	null	null	null	null	null	null	0	null	https://smeltcore.com/models/sulphur-2/
m2-max	trellis-2	TRELLIS.2-4B	3d	active	verified	17	null	null	null	null	true	null	1	trellis-2-m2-max	https://smeltcore.com/recipes/trellis-2-m2-max/
m2-max	trellis-image-large	TRELLIS image-large	3d	active	unknown	16	null	null	null	null	null	null	0	null	https://smeltcore.com/models/trellis-image-large/
m2-max	voxcpm	VoxCPM	tts	active	verified	2	null	null	null	null	true	null	1	voxcpm-m2-max	https://smeltcore.com/recipes/voxcpm-m2-max/
m2-max	voxcpm2	VoxCPM2	tts	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/voxcpm2/
m2-max	voxtral	Voxtral Mini 3B	multimodal	active	unknown	10	null	null	null	null	null	null	0	null	https://smeltcore.com/models/voxtral/
m2-max	wan-2-1	Wan 2.1	video	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/wan-2-1/
m2-max	wan-2-2	Wan 2.2 TI2V-5B	video	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/wan-2-2/
m2-max	wan-2-2-14b	Wan 2.2 14B	video	active	unknown	24	null	null	null	null	null	null	0	null	https://smeltcore.com/models/wan-2-2-14b/
m2-max	waypoint-1-5	Waypoint 1.5	3d	active	unknown	12	null	null	null	null	null	null	0	null	https://smeltcore.com/models/waypoint-1-5/
m2-max	wizardlm2-7b	wizardlm2 7b	llm	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/wizardlm2-7b/
m2-max	z-image-turbo	Z-Image Turbo	image	active	verified	6	null	null	null	null	true	null	1	z-image-turbo-m2-max	https://smeltcore.com/recipes/z-image-turbo-m2-max/
m2-pro	acestep-1-5-xl	ACE-Step 1.5 XL	music	active	unknown	8	null	null	null	null	null	null	0	null	https://smeltcore.com/models/acestep-1-5-xl/
m2-pro	agents-a1-35b-a3b	Agents-A1 35B-A3B	multimodal	active	too_big	24	null	null	null	null	null	null	0	null	https://smeltcore.com/models/agents-a1-35b-a3b/

End of preview. Expand in Data Studio

Self-Hosted AI — GPU Compatibility, Recipes and Catalogue

Which open-weight AI models actually run on which consumer GPU, and what it takes to get them running. 2 646 model×GPU verdicts across 98 models and 27 cards, plus 973 full setup guides (17 MB of markdown) written against specific hardware.

This is the machine-readable form of smeltcore.com. Every row carries a url back to the page it came from.

Generated 2026-08-31T19:42:36+00:00 from the public read API (https://api.smeltcore.com/api/v1) — no private data, no credentials, reproducible by anyone.

Configs

config	rows	what it is
`compatibility`	2 646	the point of the dataset. One row per model × GPU, with a verdict
`recipes`	973	full setup guides, markdown included, tagged by model / GPU / tool
`models`	98	the catalogue: licence, upstream repo, modality
`gpus`	27	the cards: VRAM, vendor, series
`benchmark_sources`	158	third-party measurements, normalised and cited

from datasets import load_dataset

compat = load_dataset("REPO_ID", "compatibility", split="train")
compat.filter(lambda r: r["gpu_slug"] == "rtx-4090" and r["fit"] == "verified")

The `fit` scale

The whole dataset turns on this column, so it is worth reading before using it.

verdict	rows	meaning
`verified`	972	somebody ran it on this exact card and wrote down how — there is a recipe behind the row
`fits`	659	inferred: the model's memory floor is under the card's VRAM, and the vendor is supported. Not measured
`unknown`	455	no floor established for this model, so no honest call can be made
`too_big`	560	the memory floor exceeds this card. This is the one verdict asserted from anywhere, not only from same-vendor evidence

The asymmetry between fits and too_big is deliberate: a model is only called runnable on evidence from the same vendor's hardware, but it is called not runnable from a memory floor established anywhere. Being wrong in the optimistic direction wastes somebody's evening; being wrong in the pessimistic direction only costs them a model they could have tried.

min_vram_gb is a filter floor in decimal GB — the smallest card the model is offered on — not a measured peak. Measured peaks, where they exist, are in peak_vram_gb and in benchmark_sources.

Provenance, stated plainly

recipes is first-party. Written for this catalogue against named hardware, with the quantization, runtime and settings each one was written for.

benchmark_sources is not. 111 of 158 rows come from a single third-party site (www.hardware-corner.net), and 4 were measured by us. It is published as a citation index, not as our benchmarks: what this project contributes is the normalisation — one model slug, one GPU slug, one unit convention — and every row is required to carry source_url back to whoever did the measuring. Credit and verification both belong there. If you use a number from this table, cite the source row, not this dataset.

confidence is a 0–1 score reflecting how much the source is trusted; it is not a statistical confidence interval.

Coverage and what it is not

27 consumer cards — NVIDIA, AMD and Apple silicon. No datacenter GPUs (no H100, no A100): this catalogue is about hardware people own.
8 modalities: llm (38), multimodal (18), image (14), video (14), tts (6), 3d (4), music (2), specialized (2).
Verdicts are about whether it runs, not how well it performs. There is no quality benchmark here and no leaderboard.
The catalogue moves — models get added, quantizations appear weekly. A stale copy of this dataset will understate coverage. generated_at above is the only date that matters.

Licence and attribution

Released under CC BY-SA 4.0, matching the licence the site publishes its data under. Attribution goes to smeltcore.com.

Rows in benchmark_sources describe third-party work; that licence does not extend to the measurements themselves, which belong to the sites named in source_url.

Citation

@misc{smeltcore_selfhosted_ai,
  title  = {Self-Hosted AI — GPU Compatibility, Recipes and Catalogue},
  author = {smeltcore},
  url    = {https://smeltcore.com},
  note   = {Generated 2026-08-31T19:42:36+00:00},
  year   = {2026}
}

Downloads last month: 105