Local AI memory files
Each model page separates weight footprint from runtime overhead and context reserve.
Language models
Llama 3.1 8B
chat, RAG, general assistants
8.0B parametersInspect →MetaLlama 3.3 70B
high-quality chat and reasoning
70.6B parametersInspect →QwenQwen2.5 7B
compact multilingual chat
7.6B parametersInspect →QwenQwen2.5 14B
balanced local assistant
14.7B parametersInspect →QwenQwen2.5 32B
higher-quality chat and coding
32.5B parametersInspect →QwenQwen2.5 72B
large multilingual inference
72.7B parametersInspect →QwenQwen3 8B
compact reasoning and chat
8.2B parametersInspect →QwenQwen3 14B
mid-size reasoning
14.8B parametersInspect →QwenQwen3 32B
large local reasoning
32.8B parametersInspect →DeepSeekDeepSeek-R1 Distill Qwen 7B
compact distilled reasoning
7.6B parametersInspect →DeepSeekDeepSeek-R1 Distill Qwen 14B
mid-size distilled reasoning
14.8B parametersInspect →DeepSeekDeepSeek-R1 Distill Qwen 32B
larger distilled reasoning
32.8B parametersInspect →GoogleGemma 3 4B
small multimodal assistant
4.3B parametersInspect →GoogleGemma 3 12B
mid-size multimodal assistant
12.0B parametersInspect →GoogleGemma 3 27B
larger multimodal assistant
27.0B parametersInspect →MicrosoftPhi-4 14B
compact reasoning and coding
14.0B parametersInspect →MistralMistral 7B v0.3
general local inference
7.3B parametersInspect →MistralMistral Small 24B
mid-size general assistant
24.0B parametersInspect →MistralMixtral 8x7B
mixture-of-experts inference
46.7B parametersInspect →MistralCodestral 22B
code completion and generation
22.0B parametersInspect →DeepSeekDeepSeek Coder 6.7B
local coding assistant
6.7B parametersInspect →DeepSeekDeepSeek Coder 33B
larger code generation
33.0B parametersInspect →01.AIYi 34B
general long-context use
34.0B parametersInspect →CohereCommand R 35B
RAG and tool use
35.0B parametersInspect →Image models
Stable Diffusion 1.5
local image generation
0.9B parametersInspect →Stability AIStable Diffusion XL
high-resolution image generation
6.6B parametersInspect →Black Forest LabsFLUX.1 Schnell
fast image generation
12.0B parametersInspect →Black Forest LabsFLUX.1 Dev
high-quality image generation
12.0B parametersInspect →