The best open LLMs you can actually run on a laptop

Open LLMs light enough for a real laptop — fitting in about 6 GB at their default quant, so they leave headroom for the OS on an entry laptop GPU or a 16 GB Mac's shared memory. This is a deliberately tighter budget than a desktop's full 8 GB card, ranked by the most model you can squeeze onto portable hardware, with the honest $0 offline cost.

How this is ranked: Objective fit filter at a laptop-conservative 6 GB (tighter than the 8 GB desktop page, since a laptop shares memory with the OS). 'Best for a laptop' = 'comfortably runs on portable hardware,' computed by the engine — not a quality ranking. Exact fit depends on the user's RAM/VRAM; the advisor checks their machine.

1. Qwen2.5 Math 7B Instruct — Qwen, 7.6B · ~6.0 GB VRAM · $0.16/1M API est. · commercial OK
2. OLMoE 1B 7B 0125 Instruct — allenai, 6.9B · ~6.0 GB VRAM · $0.16/1M API est. · commercial OK
3. OLMoE 1B 7B 0924 — allenai, 6.9B · ~6.0 GB VRAM · $0.16/1M API est. · commercial OK
4. Josiefied Qwen3 VL 4B Instruct abliterated beta v1 — Goekdeniz-Guelmez, 4.4B · ~6.0 GB VRAM · $0.14/1M API est. · non-commercial
5. Qwen3 4B — Qwen, 4B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
6. Qwen3 4B Instruct 2507 — Qwen, 4B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
7. Rio 3.0 Open Mini — prefeitura-rio, 4B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
8. NVIDIA Nemotron 3 Nano 4B BF16 — nvidia, 4B · ~6.0 GB VRAM · $0.13/1M API est. · non-commercial
9. Qwen3 4B Base — Qwen, 4B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
10. Qwen3 4B Thinking 2507 — Qwen, 4B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
11. Nanbeige4.1 3B — Nanbeige, 3.9B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
12. Phi-3.5-mini Instruct — Microsoft, 3.8B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
13. Phi 4 mini instruct — microsoft, 3.8B · ~6.0 GB VRAM · $0.13/1M API est. · commercial OK
14. Phi tiny MoE instruct — microsoft, 3.8B · ~4.0 GB VRAM · $0.13/1M API est. · commercial OK
15. Phi 3 mini 4k instruct — microsoft, 3.8B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
16. Nemotron Labs Diffusion 3B Base — nvidia, 3.8B · ~4.0 GB VRAM · $0.13/1M API est. · non-commercial
17. Phi 4 mini reasoning — microsoft, 3.8B · ~6.0 GB VRAM · $0.13/1M API est. · commercial OK
18. Nemotron Labs Diffusion 3B — nvidia, 3.8B · ~5.0 GB VRAM · $0.13/1M API est. · non-commercial
19. HyperCLOVAX SEED Vision Instruct 3B — naver-hyperclovax, 3.7B · ~5.0 GB VRAM · $0.13/1M API est. · non-commercial
20. PowerLM 3b — ibm-research, 3.5B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
21. PowerMoE 3b — ibm-research, 3.4B · ~4.0 GB VRAM · $0.13/1M API est. · commercial OK
22. granite 4.1 3b — ibm-granite, 3.4B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
23. granite 4.0 micro — ibm-granite, 3.4B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
24. tiny aya base — CohereLabs, 3.3B · ~5.0 GB VRAM · $0.13/1M API est. · non-commercial
25. tiny aya global — CohereLabs, 3.3B · ~5.0 GB VRAM · $0.13/1M API est. · non-commercial
26. Llama 3.2 3B — meta-llama, 3.2B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
27. Llama 3.2 3B Instruct — unsloth, 3.2B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
28. Llama 3.2 3B Instruct pythonic — baseten, 3.2B · ~5.0 GB VRAM · $0.13/1M API est. · commercial OK
29. Qwen2.5 3B Instruct — Qwen, 3.1B · ~4.0 GB VRAM · $0.12/1M API est. · non-commercial
30. Qwen2.5 Coder 3B — Qwen, 3.1B · ~4.0 GB VRAM · $0.12/1M API est. · non-commercial
31. SmolLM3 3B — HuggingFaceTB, 3.1B · ~5.0 GB VRAM · $0.12/1M API est. · commercial OK
32. Qwen2.5 3B — Qwen, 3.1B · ~4.0 GB VRAM · $0.12/1M API est. · non-commercial
33. SmolLM3 3B Base — HuggingFaceTB, 3.1B · ~5.0 GB VRAM · $0.12/1M API est. · commercial OK
34. Qwen2.5 Coder 3B Instruct — Qwen, 3.1B · ~4.0 GB VRAM · $0.12/1M API est. · non-commercial
35. Llama 3.2 3B Instruct — Meta, 3B · ~5.0 GB VRAM · $0.12/1M API est. · commercial OK
36. starcoder2 3b — bigcode, 3B · ~4.0 GB VRAM · $0.12/1M API est. · commercial OK
37. kimi k2.6 eagle3 mla — lightseekorg, 3B · ~4.0 GB VRAM · $0.12/1M API est. · non-commercial
38. phi 2 — microsoft, 2.8B · ~4.0 GB VRAM · $0.12/1M API est. · commercial OK
39. stablelm 3b 4e1t — stabilityai, 2.8B · ~5.0 GB VRAM · $0.12/1M API est. · commercial OK
40. gpt neo 2.7B — EleutherAI, 2.7B · ~4.0 GB VRAM · $0.12/1M API est. · commercial OK

Showing the top 40 of 122. See all →

More: all "best" lists · cost calculator · all models