Updated 2026-09-20 · ranked by real download data
Trending text-generation
Ranked nightly from our download snapshots of the Hugging Face catalog. Every entry shows its licence and what hardware it realistically needs.
| # | Model | Params | Context | Commercial use | 30d | Min VRAM |
|---|---|---|---|---|---|---|
| 01 | pythia-6.9b | 7.0 B | 2 K | ✓ apache-2.0 | 757 K | from 16.9 GB |
| 02 | EXAONE-3.5-7.8B-Instruct-AWQ | 7.8 B | 33 K | other | 578 K | from 18.9 GB |
| 03 | bge-reranker-v2.5-gemma2-lightweight-gptq | 9.2 B | 8 K | unknown | 534 K | from 22.2 GB |
| 04 | GLM-5.3-Flash-GGUF | — | — | ✓ mit | 630 K | — |
| 05 | GLM-5.3 | 753.3 B | 1.0 M | other | 947 K | from 1,770.8 GB |
| 06 | Qwen2.5-Math-1.5B-Instruct | 1.5 B | 4 K | ✓ apache-2.0 | 299 K | from 4.1 GB |
| 07 | NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 | 18.2 B | 262 K | other | 762 K | from 43.4 GB |
| 08 | zeta-2.1-autoround-W4A16 | 2.2 B | 33 K | unknown | 427 K | from 5.7 GB |
| 09 | granite-4.1-3b | 3.4 B | 131 K | ✓ apache-2.0 | 380 K | from 8.5 GB |
| 10 | DeepSeek-V3.2 | 685.4 B | 164 K | ✓ mit | 2.4 M | from 1,611.2 GB |
| 11 | Kimi-K3-DSpark | 2.3 B | 1.0 M | unknown | 3.6 M | from 5.8 GB |
| 12 | Ornith-1.5-9B-OBLITERATED | 9.7 B | — | ✓ mit | 359 K | from 23.2 GB |
| 13 | Ornith-1.5-9B-GGUF | — | — | ✓ mit | 5.8 M | — |
| 14 | Qwen2.5-Coder-1.5B | 1.5 B | 33 K | ✓ apache-2.0 | 349 K | from 4.1 GB |
| 15 | Qwen3-235B-A22B-Instruct-2507-FP8 | 235.1 B | 262 K | ✓ apache-2.0 | 382 K | from 553.0 GB |
| 16 | DeepSeek-V4-Flash-DSpark | 165.3 B | 1.0 M | ✓ mit | 1.0 M | from 388.9 GB |
| 17 | Ornith-1.5-397B-GGUF | — | — | ✓ mit | 1.4 M | — |
| 18 | Llama-2-7b-chat-hf | 6.7 B | — | ⚠ llama2 | 430 K | from 16.3 GB |
| 19 | openai-gpt | 120 M | — | ✓ mit | 268 K | from 0.8 GB |
| 20 | NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 | 302.8 B | 262 K | other | 345 K | from 712.2 GB |
| 21 | Olmo-3-7B-Instruct | 7.3 B | 66 K | ✓ apache-2.0 | 276 K | from 17.7 GB |
| 22 | Qwen3-Coder-480B-A35B-Instruct-FP8 | 480.2 B | 262 K | ✓ apache-2.0 | 804 K | from 1,128.9 GB |
| 23 | CodeLlama-7b-hf | 6.7 B | 16 K | ⚠ llama2 | 325 K | from 16.3 GB |
| 24 | llama-7b | 6.7 B | 2 K | other | 237 K | from 16.3 GB |
| 25 | LFM2.5-8B-A1B-GGUF | — | — | other | 563 K | — |
Membership and ranking refresh nightly after the snapshot run. VRAM is an estimate for the smallest quantization at 8K context — methodology.