Models

Top open-source models

The most-downloaded open-weight language models on Hugging Face right now - the ones actually worth running at home. What each is good at, roughly what VRAM it needs, and a shortcut to the GGUF quants. Pulled from the Hugging Face API at build time and filtered to recognised labs.

Checked 22 July 2026. Ranked by Hugging Face downloads, which leans towards small models - the most-pulled model is not the best pick for your machine.

Check what fits your GPU →

1. 01
   Qwen3-8B 8B Chat

   Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

   Downloads

   16.4M

   Likes

   1k

   VRAM

   runs on 8GB

   Model GGUF quants
2. 02
   Qwen2.5-1.5B-Instruct 1.5B Chat

   Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

   Downloads

   13.2M

   Likes

   780

   VRAM

   runs on 8GB

   Model GGUF quants
3. 03
   Qwen2.5-7B-Instruct 7B Chat

   Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

   Downloads

   12.0M

   Likes

   1k

   VRAM

   runs on 8GB

   Model GGUF quants
4. 04
   Llama-3.2-1B-Instruct 1B Chat

   Meta. The open-weight workhorse - broad, reliable, huge tooling support.

   Downloads

   10.4M

   Likes

   2k

   VRAM

   runs on 8GB

   Model GGUF quants
5. 05
   DeepSeek-R1 CodingReasoningChat

   DeepSeek. Reasoning and coding powerhouse that punches well above its weight.

   Downloads

   9.2M

   Likes

   13k

   VRAM

   Model GGUF quants
6. 06
   Qwen3.6-35B-A3B-NVFP4 3B ChatMoE

   Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

   Downloads

   8.6M

   Likes

   486

   VRAM

   runs on 8GB

   Model GGUF quants
7. 07
   Llama-3.1-8B-Instruct 8B Chat

   Meta. The open-weight workhorse - broad, reliable, huge tooling support.

   Downloads

   8.4M

   Likes

   6k

   VRAM

   runs on 8GB

   Model GGUF quants
8. 08
   Qwen3-32B 32B Chat

   Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

   Downloads

   8.3M

   Likes

   719

   VRAM

   needs a 24GB card

   Model GGUF quants
9. 09
   gpt-oss-20b 20B Chat

   OpenAI. OpenAI's open-weight release - strong general chat and reasoning.

   Downloads

   7.3M

   Likes

   5k

   VRAM

   needs ~12GB

   Model GGUF quants
10. 10
    Qwen2.5-3B-Instruct 3B Chat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    5.9M

    Likes

    540

    VRAM

    runs on 8GB

    Model GGUF quants
11. 11
    Qwen3-1.7B 1.7B Chat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    5.8M

    Likes

    503

    VRAM

    runs on 8GB

    Model GGUF quants
12. 12
    Qwen3-4B 4B Chat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    5.1M

    Likes

    660

    VRAM

    runs on 8GB

    Model GGUF quants
13. 13
    Qwen3-14B 14B Chat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    4.5M

    Likes

    430

    VRAM

    needs ~12GB

    Model GGUF quants
14. 14
    gpt-oss-120b 120B Chat

    OpenAI. OpenAI's open-weight release - strong general chat and reasoning.

    Downloads

    4.4M

    Likes

    5k

    VRAM

    big-memory only

    Model GGUF quants
15. 15
    gemma-3-1b-it 1B Chat

    Google. Compact and capable - a lot of quality in a small footprint.

    Downloads

    4.0M

    Likes

    1k

    VRAM

    runs on 8GB

    Model GGUF quants
16. 16
    Qwen2-1.5B-Instruct 1.5B Chat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    3.8M

    Likes

    164

    VRAM

    runs on 8GB

    Model GGUF quants
17. 17
    Qwen2.5-14B-Instruct 14B Chat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    3.3M

    Likes

    354

    VRAM

    needs ~12GB

    Model GGUF quants
18. 18
    Qwen3-30B-A3B 3B ChatMoE

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    3.1M

    Likes

    910

    VRAM

    runs on 8GB

    Model GGUF quants
19. 19
    Qwen2.5-Coder-14B-Instruct 14B CodingChat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    3.1M

    Likes

    174

    VRAM

    needs ~12GB

    Model GGUF quants
20. 20
    GLM-5.2-FP8 ChatMoE

    Zhipu. Strong bilingual chat and agentic use.

    Downloads

    3.0M

    Likes

    227

    VRAM

    Model GGUF quants
21. 21
    DeepSeek-V4-Flash Chat

    DeepSeek. Reasoning and coding powerhouse that punches well above its weight.

    Downloads

    3.0M

    Likes

    2k

    VRAM

    Model GGUF quants
22. 22
    Qwen2.5-32B-Instruct 32B Chat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    2.9M

    Likes

    355

    VRAM

    needs a 24GB card

    Model GGUF quants
23. 23
    Qwen3-Coder-Next-FP8 CodingChat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    2.8M

    Likes

    162

    VRAM

    Model GGUF quants
24. 24
    Gemma-4-31B-IT-NVFP4 31B Chat

    Google. Compact and capable - a lot of quality in a small footprint.

    Downloads

    2.5M

    Likes

    539

    VRAM

    needs a 24GB card

    Model GGUF quants
25. 25
    TinyLlama-1.1B-Chat-v1.0 1.1B CodingChat

    Meta. The open-weight workhorse - broad, reliable, huge tooling support.

    Downloads

    2.5M

    Likes

    2k

    VRAM

    runs on 8GB

    Model GGUF quants
26. 26
    Ornith-1.0-9B-GGUF 9B Chat

    Open-weight text-generation model.

    Downloads

    2.5M

    Likes

    546

    VRAM

    runs on 8GB

    Model GGUF quants
27. 27
    granite-4.1-8b 8B Chat

    Open-weight text-generation model.

    Downloads

    2.4M

    Likes

    225

    VRAM

    runs on 8GB

    Model GGUF quants
28. 28
    deepseek-v4-gguf MoE

    DeepSeek. Reasoning and coding powerhouse that punches well above its weight.

    Downloads

    2.4M

    Likes

    335

    VRAM

    Model GGUF quants
29. 29
    NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4 12B CodingChatMoE

    Open-weight text-generation model.

    Downloads

    2.3M

    Likes

    393

    VRAM

    runs on 8GB

    Model GGUF quants
30. 30
    DeepSeek-R1-0528-Qwen3-8B 8B ReasoningChat

    Alibaba. Superb all-rounder - strong at coding, maths and multilingual.

    Downloads

    2.3M

    Likes

    1k

    VRAM

    runs on 8GB

    Model GGUF quants

Ranked by downloads on the Hugging Face Hub. Model sizes and VRAM figures are rough guides - use the calculator for your exact card and quant.