Family index · 9 families

Local LLM families

Most open models ship in a ladder of sizes. Pick a family to see every size side by side by memory, then choose the largest that fits your hardware. Validated 2026-06-15.

Qwen3

8 sizes · 0.6B to 235B
Gemma

7 sizes · 1B to 27B
Qwen2.5

7 sizes · 0.494B to 72B
Qwen2.5-Coder

6 sizes · 0.494B to 32B
Llama

4 sizes · 1B to 70B
Mistral

4 sizes · 7B to 46.7B
DeepSeek-R1-Distill

4 sizes · 7B to 32B
S

Sarvam

4 sizes · 2B to 105B
SmolLM2

3 sizes · 0.135B to 1.7B

Prefer to start from your hardware? Browse by memory →