{"slug": "downloadsalltime-decreasing-day-over-day-for-multiple-public-models-expected", "title": "downloadsAllTime decreasing day-over-day for multiple public models — expected behavior?", "summary": "A developer reports that the downloadsAllTime field for multiple public models on Hugging Face Hub is decreasing day-over-day, contrary to expectations for a cumulative count. The observed drops range from 3.6% to 10.4% over a five-day period for models like Qwen/Qwen3.5-4B and deepseek-ai/DeepSeek-V4-Pro. The developer seeks clarification on whether this is expected behavior or indicates retroactive adjustments, and requests guidance on computing reliable download trends.", "body_md": "Hi Hugging Face team,\n\nI built a program to continuously monitor download counts for popular public models across platforms. On the Hugging Face Hub, it queries the API using `downloadsAllTime`\n\n(not the 30-day `downloads`\n\nfield) and stores one snapshot per model per day.\n\nRecently I noticed that ** downloadsAllTime can decrease** when queried on consecutive days. Since this field is documented as the cumulative download count since repo creation, I wanted to check whether this is expected — or whether it might indicate a retroactive adjustment on the Hub side.\n\n``` python\nfrom huggingface_hub import model_info\n\ninfo = model_info(\"Qwen/Qwen3.5-4B\", expand=[\"downloadsAllTime\"])\nprint(info.downloads_all_time)\n```\n\nI always pass `expand=[\"downloadsAllTime\"]`\n\n, as recommended in [this thread](https://discuss.huggingface.co/t/model-download-statistics/145580).\n\nBelow are daily snapshots from my program (same Hub API query each day). These are all relatively new releases (Qwen 3.5 series and DeepSeek V4), yet they already show the same pattern:\n\n**1. Qwen/Qwen3.5-4B**\n\n| Date | `downloadsAllTime` |\n|---|---|\n| 2026-06-08 | 19,704,972 |\n| 2026-06-09 | 19,329,213 |\n| 2026-06-13 | 18,001,904 |\n\nCurrent value (queried on 2026-06-13): **18,001,904**\n\nChange 2026-06-08 → 2026-06-13: **−1,703,068 (−8.6%)**\n\n**2. Qwen/Qwen3.5-35B-A3B**\n\n| Date | `downloadsAllTime` |\n|---|---|\n| 2026-06-08 | 11,183,510 |\n| 2026-06-09 | 11,018,594 |\n| 2026-06-13 | 10,782,371 |\n\nCurrent value (queried on 2026-06-13): **10,782,371**\n\nChange 2026-06-08 → 2026-06-13: **−401,139 (−3.6%)**\n\n**3. deepseek-ai/DeepSeek-V4-Pro**\n\n| Date | `downloadsAllTime` |\n|---|---|\n| 2026-06-08 | 6,692,783 |\n| 2026-06-09 | 6,288,488 |\n| 2026-06-13 | 5,999,035 |\n\nCurrent value (queried on 2026-06-13): **5,999,035**\n\nChange 2026-06-08 → 2026-06-13: **−693,748 (−10.4%)**\n\nAll three models show a monotonic decrease across these snapshots. The repos themselves were not deleted or renamed during this period.\n\nThe program is meant to track how popular models are growing over time across the ecosystem. For Hugging Face, it computes day-over-day differences of `downloadsAllTime`\n\nto estimate daily new downloads and plot trends.\n\nLately, several models show **negative daily increments** because the cumulative total itself went down between snapshots. That makes the charts hard to interpret — readers naturally expect download counts to only increase.\n\nI understand that the `downloads`\n\nfield (30-day rolling window) can decrease, but I had assumed `downloadsAllTime`\n\nwas non-decreasing unless there was an explicit data correction. I’m trying to figure out whether I should change how my program handles Hugging Face stats, or whether this reflects something on the Hub side.\n\n**Is it expected that downloadsAllTime can decrease over time?** If so, under what circumstances (deduplication, bot filtering, bug fixes, counting rule changes, etc.)?\n\n**Has there been a recent retroactive adjustment** on the Hub that could explain broad downward shifts across many public models in early/mid June 2026?\n\n**Is there a recommended way** for downstream consumers to compute reliable daily download trends from the public API, given that all-time totals may be revised?\n\nHappy to share more model IDs, raw API responses, or a minimal reproduction script if that would help investigate.\n\nThanks!", "url": "https://wpnews.pro/news/downloadsalltime-decreasing-day-over-day-for-multiple-public-models-expected", "canonical_source": "https://discuss.huggingface.co/t/downloadsalltime-decreasing-day-over-day-for-multiple-public-models-expected-behavior/176918#post_1", "published_at": "2026-06-18 02:52:27+00:00", "updated_at": "2026-06-18 02:58:40.018267+00:00", "lang": "en", "topics": ["ai-tools", "ai-infrastructure", "developer-tools"], "entities": ["Hugging Face", "Qwen", "DeepSeek", "Qwen3.5-4B", "Qwen3.5-35B-A3B", "DeepSeek-V4-Pro"], "alternates": {"html": "https://wpnews.pro/news/downloadsalltime-decreasing-day-over-day-for-multiple-public-models-expected", "markdown": "https://wpnews.pro/news/downloadsalltime-decreasing-day-over-day-for-multiple-public-models-expected.md", "text": "https://wpnews.pro/news/downloadsalltime-decreasing-day-over-day-for-multiple-public-models-expected.txt", "jsonld": "https://wpnews.pro/news/downloadsalltime-decreasing-day-over-day-for-multiple-public-models-expected.jsonld"}}