Updates and Freshness

What we track, how often we check it, and where the next gaps still are.

This is the public operations board for the hub. It shows the current data categories, their refresh cadence, the types of sources we rely on, and the last few changes visible in each lane. Freshness matters, but official verification still comes first.

8

Tracked categories

3

Fully automated lanes

4

Mixed automation lanes

1

Manual editorial lanes

Current operating model

The hub now refreshes core data throughout the day, but not every category should be treated the same. Prices, status pages, and structured benchmark feeds are strong candidates for hourly automation. Social signals and editorial reactions should trigger review work, not silently overwrite the canonical data layer.

Activity log News pipeline Collection policy References

Manual social watch

Social can tell us when to look, not what is true.

X / Twitter is useful for launch timing and first impressions, but it should stay manual-review only or official API only. The automated layer should still rely on official docs, pricing pages, changelogs, and status endpoints.

Latest refresh 14 Apr 2026, 22:00

AI milestones

Mixed Watch gap

Keep major launch dates, lab founding dates, and historical AI milestones usable without faking precision where the sourcing is weak.

Cadence

Curated updates as research lands

Last refreshed

14 Apr 2026, 22:00

1 min ago

Source types

  • Official launch posts, model cards, and company pages
  • Official registry and filing sources where available
  • Selective secondary reporting only when official chronology is weak

Estimated anchors stay in tracking notes, not in the canonical public date fields.

Last 3 visible updates

Open category
1 min ago

Llama 3 released

Next exact anniversary in 4 days.
1 min ago

Moonshot AI founded

Still marked tracking until stronger chronology evidence is sourced.

Regional availability

Mixed Manual review

Track where models, apps, and API surfaces are officially available so launch coverage does not assume every market has access.

Cadence

Manual review with official-source refresh

Last refreshed

09 Apr 2026, 20:49

5 days ago

Source types

  • Official support-country and region-availability pages
  • Product FAQs and help-center coverage for country restrictions
  • Official docs and product availability notices

This lane is intentionally official-first and cautious. Unknown availability should stay unknown until the provider documents it.

Last 3 visible updates

Open category
5 days ago

Sora app and Sora 2

Model-specific rule for 2 tracked models on Web + App.
5 days ago

Claude supported locations

Provider baseline rule for 14 tracked models on Web + App.
5 days ago

Google AI Studio and Gemini API

Provider baseline rule for 19 tracked models on API + Web.

Models and releases

Mixed Healthy

Watch new model names, release desk entries, and tracked provider catalog changes as close to launch as possible.

Cadence

Hourly automated refresh

Last refreshed

14 Apr 2026, 21:06

54 min ago

Source types

  • Official provider docs, changelogs, and model pages
  • OpenRouter model discovery and routed catalog deltas
  • Manual social watch for launch timing, then official-source verification

Social posts are a trigger, not the source of truth. We should verify launches against official docs before the public data layer changes.

Last 3 visible updates

Open category
7 days ago

Claude Opus 4.6 (Fast)

Anthropic release desk entry is ready for editor with 2 related stories and 0 benchmark signals attached.
7 days ago

GLM 5.1

Zhipu AI release desk entry is needs research with 0 related stories and 0 benchmark signals attached.
7 days ago

Claude Mythos Preview

Anthropic release desk entry is ready for editor with 4 related stories and 0 benchmark signals attached.

Benchmarks and evals

Automated Healthy

Keep the benchmark library and weighted score layer fresh enough that new releases can be compared quickly and honestly.

Cadence

Hourly automated refresh

Last refreshed

14 Apr 2026, 21:06

54 min ago

Source types

  • Artificial Analysis and other benchmark owners
  • LMSYS / arena-style public benchmark sources
  • Research repositories and public eval feeds where intended for reuse

Last 3 visible updates

Open category
54 min ago

Arena and headline eval track refresh

Updated the frontier conversation benchmark used in the ranking layer.
54 min ago

Quality score recompute

Rebuilt the weighted scoring layer used across the hub.

Pricing and value

Automated Healthy

Track provider pricing and value signals from official pages first, with routed API pricing used as the fastest discovery layer.

Cadence

Hourly automated refresh

Last refreshed

14 Apr 2026, 21:06

54 min ago

Source types

  • Official provider pricing pages
  • Provider APIs and model catalogs where pricing is exposed
  • OpenRouter price index for discovery and cross-checking

Last 3 visible updates

Open category
54 min ago

Official provider pricing cross-check

Validated tracked price rows against official pricing pages.
54 min ago

OpenRouter pricing refresh

Pulled live routed endpoint pricing and model discovery data.
54 min ago

Pricing validator pass

Ran the secondary sanity-check layer over live pricing snapshots.

News and release watch

Mixed Watch gap

Use AI news to catch launches, benchmarks, evals, and pricing chatter quickly, but do not let it outrank official provider sources.

Cadence

Hourly automated refresh plus manual newsroom watch

Last refreshed

14 Apr 2026, 21:06

54 min ago

Source types

  • Official blogs, newsroom pages, and public AI-focused feeds
  • Shared website news pipeline routing rules
  • Manual social monitoring where terms permit or official API access exists

X / Twitter should stay manual-review only or official API only. It can tell us when to look, but it should not be an unauthorised automated source.

Last 3 visible updates

Open category
4 hr ago

Anthropic co-founder confirms the company briefed the Trump administration on Mythos

TechCrunch (industry)
4 hr ago

The attacks on Sam Altman are a warning for the AI world

The Verge (industry)
5 hr ago

Max Hodak’s Science Corp. is preparing to place its first sensor in a human brain

TechCrunch (industry)

Provider status

Automated Healthy

Keep official provider status pages in one place so outages and degraded APIs are visible without opening every lab dashboard manually.

Cadence

Hourly automated refresh

Last refreshed

14 Apr 2026, 21:34

26 min ago

Source types

  • Official status page APIs, RSS feeds, and incident pages
  • Provider-specific status pages linked from the hub

Last 3 visible updates

Open category
26 min ago

Anthropic

All Systems Operational
26 min ago

Cohere

All systems operational
26 min ago

DeepSeek

All Systems Operational

Guides and learning

Manual Manual review

Learning pages still need a human editorial pass, especially when model launches change local-running advice, pricing, or terminology.

Cadence

Manual editorial review

Last refreshed

14 Apr 2026, 22:00

1 min ago

Source types

  • Internal guide pages and academy content
  • Official provider docs used during refreshes
  • Release-desk drafts queued for editor review

This is the least automated part of the site today, so users should expect an editorial review rhythm rather than a live ticker.

Last 3 visible updates

Open category
1 min ago

Advanced Prompting

Guide or learning page updated in the repository.
1 min ago

Ai For Research

Guide or learning page updated in the repository.
1 min ago

Ai For Writing

Guide or learning page updated in the repository.