Skip to the page
CoinjureNo wallet

New on Hugging Face's trending models: tencent/Hy4-flash at #23, a model that reads images and text and answers in text.

First on record. The first time tencent/Hy4-flash appears on this study's record, which starts Oct 31, 2026.Coinjure recorded this at 07:20 UTC on Nov 18
Finding23 hours agoaihugging-facemodel
Stamped 2026-11-18 07:20 UTC. Never edited, never back-dated.
Details, as recorded
As recorded: New on Hugging Face's trending models: tencent/Hy4-flash at #23, a model that reads images and text and answers in text
A model that reads images and text and answers in text. Its card says: "Hy4 flash is our most cost-effective Hy model to date, featuring 220B total parameters (including 64B Engram parameters) with just 12B active parameters per token." 93 likes, published in the last day. https://huggingface.co/tencent/Hy4-flash. Source: the Hugging Face Hub API, models and Spaces sorted by trendingScore (Hugging Face's own measure of recent likes). Adult content and quantised re-uploads of another model are left out. Descriptions are the authors' own words or Hugging Face's tags, not a review.
Value: 23 trending rank
From New AI tools climbing fastest, run by Coinjure. 98 findings on the record.
Data: Hugging Face Hub API, trending models and Spaces ↗, via Coinjure
Post on XThe card

Proof it was never edited

Content hash630f598c716498652955a45e28787c3281b801bf9eadd8eb4845fa84cb2bfcbb
Chain0635f73d17096a85d9a8ab98d01d15063c745d322253961920afde81a64b5d0b
How to check this yourself
  1. The hash is sha256 of this exact line (the circle, the finding's id, its stamp in milliseconds, title, body, value and unit):
    ["coinjure/finding/v1","ai-tools-climbing-fastest","kn722ce5w0tc8zqmshs0pm89fh8jfrtz",1794986406016,"New on Hugging Face's trending models: tencent/Hy4-flash at #23, a model that reads images and text and answers in text","A model that reads images and text and answers in text. Its card says: \"Hy4 flash is our most cost-effective Hy model to date, featuring 220B total parameters (including 64B Engram parameters) with just 12B active parameters per token.\" 93 likes, published in the last day. https://huggingface.co/tencent/Hy4-flash. Source: the Hugging Face Hub API, models and Spaces sorted by trendingScore (Hugging Face's own measure of recent likes). Adult content and quantised re-uploads of another model are left out. Descriptions are the authors' own words or Hugging Face's tags, not a review.",23,"trending rank"]
  2. On a Mac or Linux: printf '%s' '["coinjure/finding/v1","ai-tools-climbing-fastest","kn722ce5w0tc8zqmshs0pm89fh8jfrtz",1794986406016,"New on Hugging Face'\''s trending models: tencent/Hy4-flash at #23, a model that reads images and text and answers in text","A model that reads images and text and answers in text. Its card says: \"Hy4 flash is our most cost-effective Hy model to date, featuring 220B total parameters (including 64B Engram parameters) with just 12B active parameters per token.\" 93 likes, published in the last day. https://huggingface.co/tencent/Hy4-flash. Source: the Hugging Face Hub API, models and Spaces sorted by trendingScore (Hugging Face'\''s own measure of recent likes). Adult content and quantised re-uploads of another model are left out. Descriptions are the authors'\'' own words or Hugging Face'\''s tags, not a review.",23,"trending rank"]' | shasum -a 256
  3. The chain value is sha256 of the chain before it, a colon, and this hash. The first finding of a circle chains from 64 zeros. Change one word anywhere and every chain value after it changes.
  4. Each day Coinjure publishes one root over every circle's newest chain value, in the open API and the circle's RSS feed. Keep a copy, and any later change shows.

The proof lives on Coinjure and in every copy people keep. Writing the daily root to a public blockchain as well is a later option; it costs a small fee per day.

Which new AI models and apps are climbing fastest this week?Watch the circle and its next findings come to you. No wallet needed.
Open the circle
Build on this

Turn this into something

This knowledge is open to everyone. Use it to make something people need.

Where next