Daily download tracking since 28 Jul 2026
Which AI models are actually winning — measured daily, not guessed.
Every night we snapshot downloads, likes and file inventories for the top 1,226 open-weights models on Hugging Face at 04:00 UTC, and prices, context windows and benchmark indices for 200 API-only models via OpenRouter. That history powers trend charts, VRAM estimates, API cost and side-by-side comparisons you can't get from a model card.
Every model is labelled: open weights you can self-host, or API-only (GPT, Claude, Gemini, Grok…).
Catalog downloads · trailing 30d
1.6 B
▼ 9.4% vs previous 30d
last 60 days
40,720
compare pages
1,426
models tracked
89
days tracked
| # | Model | 30d | Δ 7d | Trend |
|---|---|---|---|---|
| 01 | Krea-2-Turbo | 169 K | ▲ 412.1% | |
| 02 | Qwen-Image-2.1-Lightning | 191 K | ▲ 392.6% | |
| 03 | snowflake-arctic-embed-l-v2.0 | 2.5 M | ▲ 95.9% | |
| 04 | gemma-4-31B-it-qat-NVFP4-Blackwell | 371 K | ▲ 84.1% | |
| 05 | MeinaMix | 38 K | ▲ 82.0% | |
| 06 | Unlimited-OCR | 2.3 M | ▲ 80.6% | |
| 07 | opus-mt-zh-en | 390 K | ▲ 73.3% | |
| 08 | Qwen-Image-2.1-Turbo-GGUF | 495 K | ▲ 68.2% |
Best value via API
AA intelligence index per blended $ (3:1 input:output, per 1M tokens)| # | Model | AA idx | Blended /1M | Value |
|---|---|---|---|---|
| 01 | Ling-3.0-flash | 20.1 | $0.032 | 638 |
| 02 | DeepSeek-V4-Flash-0731 | 34.3 | $0.067 | 515 |
| 03 | gpt-oss-20b | 10 | $0.036 | 278 |
| 04 | Ling-3.0-flash-VL | 24.6 | $0.09 | 273 |
| 05 | MiMo-V2.6-Flash-RL | 37.9 | $0.18 | 217 |
| 06 | DeepSeek-V4-Flash | 24.4 | $0.11 | 214 |
| 07 | GLM-5.3-Flash | 41.8 | $0.24 | 176 |
| 08 | MiniMax-M3.1-Flash | 42.7 | $0.26 | 163 |
| 09 | gemma-4-26B-A4B-it | 16.7 | $0.11 | 156 |
| 10 | DeepSeek-V4.1-Flash-1017 | 39.5 | $0.26 | 150 |
| # | Model | AA idx | Blended /1M | Value |
|---|---|---|---|---|
| 01 | Ling 3.1 Flash | 41.1 | $0.11 | 365 |
| 02 | Solar Mini 4 | 24.1 | $0.088 | 275 |
| 03 | Claude Haiku 5.5 | 43.4 | $0.20 | 217 |
| 04 | GPT-6 Luna | 38.1 | $0.20 | 191 |
| 05 | Mercury 2.5 | 12.3 | $0.068 | 182 |
| 06 | Nemotron 3.5 Super VL | 26.5 | $0.24 | 112 |
| 07 | Solar Pro 4 | 28.2 | $0.26 | 107 |
| 08 | Nova Micro 1.0 | 5.9 | $0.061 | 96 |
| 09 | GPT-5 Nano | 13 | $0.14 | 95 |
| 10 | GPT-5.6 Luna | 37.3 | $0.45 | 83 |
New this week
moonshotai · 23 Oct · API-only
Kimi K3.1 Pro
$2.00 in / $10.00 out per 1M · 1M ctx
RecursiveSI · 21 Oct · open weights
flint
Free in / Free out per 1M · 262.1K ctx
MiniMaxAI · 19 Oct · open weights
MiniMax-M3.1-Flash
$0.15 in / $0.60 out per 1M · 1M ctx
Fresh comparisons
Browse by hardware
For machines
Read-only JSON, no key, 60 req/min. Snapshots also ship as a daily CSV dump.
Collections
Data from the public Hugging Face and OpenRouter APIs, snapshotted daily; benchmark indices from Artificial Analysis. VRAM figures are estimates — see /methodology.