Future Index

Topics

Open-weight models

Attention around Open-weight models is about usual, and has fallen over the last 24 hours.

AIGLOBAL

Signals

What moved

The largest movements across forecasts, attention, discussion sentiment and reporting in the last 24 hours, each compared with how much that topic usually moves.

Nothing moved beyond its usual range over this window. 7 measurements were compared.

7 changes were compared.

Across signals

No relationships between signals were detected this week. Checked 1 topic: 0 had a forecast contract moving well outside its usual range and 1 had unusually high public attention, and none of them showed a second layer doing something measurable at the same time.

What was reported

Occurrences reported on this topic, the most widely reported first. One entry per occurrence, however many outlets carried it, and the count on each is the outlets that reported it independently.

  1. Cohere says Alibaba’s Qwen model tried to send a file to Beijing opens in a new tab

    1 independent origin · low confidence

    TechCrunch

What people are discussing

Posts and shared links on the discussion platforms this site reads that are about this topic, over the last two days, the most discussed first. Titles are the platform's own. Counts are the platform's where it publishes them, and otherwise the comments or shares this site read; none of it is a figure this site calculates.

  1. A Working Nintendo in ~1GB of Qwen Weights opens in a new tab

    82 comments95 pointsHacker Newsdiscussion opens in a new tab

  2. Paseo – Run Claude Code, Codex, Copilot, OpenCode from anywhere opens in a new tab

    shared in 1 postMastodon

  3. Alibaba’s Qwen AI model to integrate into Apple Intelligence in landmark partnership-Xinhua opens in a new tab

    shared in 1 postMastodon

7 more posts — show them
  1. From potato capital to token city: Ulanqab rises as China's AI hub opens in a new tab

    shared in 1 postMastodon

  2. Apple to launch iPhone AI features in China in December via Alibaba partnership opens in a new tab

    shared in 1 postMastodon

  3. Open-weight model GLM 5.3 built full Firefox exploit chains with zero human input in under 48 hours opens in a new tab

    shared in 1 postMastodon

  4. One AI rule book: chip expert urges shared standards across China, US models opens in a new tab

    shared in 1 postMastodon

  5. DeepSeek Website Code Points to Cyber, a Dedicated Cybersecurity Model in Limited Testing opens in a new tab

    shared in 1 postMastodon

  6. Chinese hackers used DeepSeek and other AI tools in sweeping global cyberattacks, report finds opens in a new tab

    shared in 1 postMastodon

  7. DeepSeek put the architecture for their always-on autonomous agent on GitHub opens in a new tab

    0 comments5 pointsHacker Newsdiscussion opens in a new tab

Forecast contracts

No forecast contract has been linked to this topic yet.

Readers' discussion on this site

Comments from readers of this site, newest first. They are opinions, not a measurement: no figure on this page is computed from them.

Thank you — a moderator will look at it.

A comment is 1 to 2,000 characters.

Log in to do that.

That topic does not accept comments.

That comment no longer exists.

Too many attempts in a short time. Wait a little and try again.

Writing is paused for a moment. Try again in a few minutes.

The site could not complete that just now. Try again in a few minutes.

Plain text, up to 2,000 characters. Links are not made clickable. Posting needs an account; you will be asked to log in. Comment rules

No comments yet.

Sources

3 contributing sources — view breakdown

Attention

  • Hacker NewsPublic discussion75.6
  • Future Index internal search and pageviewsSearch & activity50.0
  • Wikimedia Pageviews (Analytics API)Search & activity34.5

How the attention figure is calculated · method version 3.0.0 · What these figures do not say