FREE · PRIVATE · NO SIGN-UP

Local AI tools for decisions you can trust.

Plan memory, find models that fit and compare options. Your hardware choices stay in your browser.

Run the full device check →Compare models ↓
NEW · FULL HARDWARE CHECK

RAM is only one part of the answer.

Check your operating system, CPU, graphics hardware, VRAM, task and context size. Get a model shortlist and a starting inference engine without false speed claims.

Check this computer →
RAM + MODEL SIZE CALCULATOR

Will this model fit in memory?

Estimate the memory needed for model weights, runtime overhead and operating-system headroom. This is a safe starting estimate, not a speed promise.

SAFE TO TEST8 GB

Estimated safe system RAM floor

~4.5 GB model weights ~1.5 GB runtime allowance 2 GB OS headroom

The estimate uses parameter count and quantization as a planning shortcut. Real files, context length and runtime memory vary. Confirm the exact model file before installing.

HARDWARE SHORTLIST

Models that pass your RAM limit

MamiLens excludes models above its conservative memory floor. Choose a task on the main adviser before treating any result as a recommendation.

57 profiles fit 16 GB RAMRun the full compatibility check →
StarCoder2 15BBigCode · Coding16 GB floorQwen 2.5 14BAlibaba Cloud · General · Writing · Coding16 GB floorPhi-4 14BMicrosoft · General · Reasoning · Math16 GB floorQwen 3 14BAlibaba Cloud · General · Coding · Reasoning16 GB floorDeepSeek R1 14BDeepSeek · Reasoning · Math · Coding16 GB floorQwen 2.5 Coder 14BAlibaba Cloud · Coding · Tools16 GB floorCode Llama 13BMeta · Coding14 GB floorMistral Nemo 12BMistral AI · General · Writing · Tools14 GB floor
MODEL COMPARISON

Compare two proven profiles

Compare memory, size, inputs, tasks and licence. “Larger” does not automatically mean “better for your job.”

VS
Meta

Llama 3.2 1B

Llama 3.2 1B is a curated dense profile for general, writing, multilingual workloads.

Safe RAM
8 GB
Download
~1.3 GB
Inputs
Text
Licence
Llama 3.2 Community
Open verified profile →
Meta

Llama 3.1 8B

Llama 3.1 8B is a curated dense profile for general, writing, coding workloads.

Safe RAM
11 GB
Download
~4.9 GB
Inputs
Text
Licence
Llama 3.1 Community
Open verified profile →
HOW TO USE THESE RESULTS

Fit first. Benchmark second.

Memory fit is only the first test

A model can load and still respond too slowly for your work.

Context uses more memory

Long conversations and large documents increase memory use.

GPU support depends on your engine

VRAM alone does not prove that a runtime supports your exact graphics hardware.