Uploaded September 2026 | Updated September 2026, 3 hours ago
Today in AI news for Wednesday, September 23, 2026: Anthropic's Claude Opus 5.5 pricing, QuiverAI's Arrow2 editable vector artwork, the Compositor layer editor for Mac, and the ZCode code-upload reports.
Claude Opus 5.5 figures (prices, speed, the HAProxy rewrite and the research-report test) are Anthropic's own announcement and internal results, not independent tests. Prices are standard API rates in USD per million tokens.
ZCode: reporting by Reuters (Sep 21, 2026). The fix, the open-sourcing and the deletion assessment are Z.ai's account and have not been independently confirmed.
The flower, layer and Photoshop-import animations are explanatory illustrations, not output captured from Arrow2 or Compositor.
Chapters:
0:00 Headlines
0:12 Claude Opus 5.5: prices, speed, Anthropic’s tests and limits
1:23 Arrow2 (QuiverAI): editable vector art
1:39 Compositor: free open-source Mac layer editor
2:04 ZCode: reported code uploads and Z.ai’s response
Today in AI news for Wednesday, September 23, 2026: Anthropic's Claude Opus 5.5 pricing, QuiverAI's Arrow2 editable vector artwork, the Compositor layer editor for Mac, and the ZCode code-upload reports.
Claude Opus 5.5 figures (prices, speed, the HAProxy rewrite and the research-report test) are Anthropic's own announcement and internal results, not independent tests. Prices are standard API rates in USD per million tokens.
ZCode: reporting by Reuters (Sep 21, 2026). The fix, the open-sourcing and the deletion assessment are Z.ai's account and have not been independently confirmed.
The flower, layer and Photoshop-import animations are explanatory illustrations, not output captured from Arrow2 or Compositor.
Chapters:
0:00 Headlines
0:12 Claude Opus 5.5: prices, speed, Anthropic’s tests and limits
1:23 Arrow2 (QuiverAI): editable vector art
1:39 Compositor: free open-source Mac layer editor
2:04 ZCode: reported code uploads and Z.ai’s response







![[SUPERSEDED DRAFT] Claude Project Memory: 3 Simple Layers
Anthropic describes different levels in how Claude thinks. That prompted a practical project-memory analogy: give scratch work, current project decisions, and stable core rules separate layers—and always back up before cleanup.
This layered-memory workflow is my analogy and recommendation, not an Anthropic research claim.
Original Anthropic video: https://www.youtube.com/watch?v=rKV5JcALQoQ
The workflow:
0. Back up
1. Scratch: todays temporary work
2. Project: current decisions and open tasks
3. Core: stable rules and architecture
#ClaudeAI #AICoding #DeveloperTools [SUPERSEDED DRAFT] Claude Project Memory: 3 Simple Layers](https://i.ytimg.com/vi/aCQWlJSCtts/mqdefault.jpg)


