Local AI Models: RAM and Hardware Finder
LOCAL AI MONITORWhich AI model will run
Which AI model will run
on your computer
Compare memory requirements, model size, license and popularity without sending your data to a cloud.
DATA UPDATE14.09.2026 20:18Hugging Face statistics · hourly
Your available memory
Alibaba Qwen8B
Qwen3 8B
Универсальная локальная модель
≈ 6 GB RAM in Q432K context
- License
- Apache 2.0
- Runs with
- Ollama · LM Studio · llama.cpp
- Downloads
- 12.7M
- Likes
- 2.0K
Google12B
Gemma 3 12B
Текст и изображения
≈ 9 GB RAM in Q4128K context
- License
- Gemma
- Runs with
- Ollama · LM Studio · llama.cpp
- Downloads
- 681.2K
- Likes
- 828
Microsoft4B
Phi-4 Mini
Слабые ПК и быстрые ответы
≈ 4 GB RAM in Q4128K context
- License
- MIT
- Runs with
- Ollama · ONNX · llama.cpp
- Downloads
- 401.9K
- Likes
- 829
DeepSeek14B
DeepSeek R1 Distill 14B
Рассуждения и код
≈ 10 GB RAM in Q464K context
- License
- MIT
- Runs with
- Ollama · LM Studio · llama.cpp
- Downloads
- 340.6K
- Likes
- 682
Mistral AI24B
Mistral Small 3.1 24B
Сильная универсальная модель
≈ 18 GB RAM in Q4128K context
- License
- Apache 2.0
- Runs with
- Ollama · vLLM · llama.cpp
- Downloads
- 405.8K
- Likes
- 1.4K
Alibaba Qwen32B
Qwen3 32B
Код и сложные инструкции
≈ 22 GB RAM in Q432K context
- License
- Apache 2.0
- Runs with
- Ollama · LM Studio · vLLM
- Downloads
- 4.7M
- Likes
- 747
Meta70B
Llama 3.3 70B
Серверный локальный ассистент
≈ 48 GB RAM in Q4128K context
- License
- Llama
- Runs with
- Ollama · vLLM · llama.cpp
- Downloads
- 919.8K
- Likes
- 3.0K
DeepSeek70B
DeepSeek R1 Distill 70B
Сложные рассуждения
≈ 48 GB RAM in Q4128K context
- License
- MIT
- Runs with
- Ollama · vLLM · llama.cpp
- Downloads
- 78.4K
- Likes
- 803
What “local AI” means
The model runs on your own computer or server. Prompts can remain on the device, internet access is optional after downloading the files, and there is no per-token API fee. RAM estimates are practical Q4 estimates and may vary by runtime, context and offloading.