2026 · 09 · 08 · Tue

Updates for 9/8

This morning: Claude's flagship shifts to Fable 5.1, and Copilot starts juggling multiple models. Plus 6 usage tips today.

Today's moves

3

Model, pricing and availability changes we actually wrote into the map and the encyclopedia.

A · Theme of the day

Claude's Top Tier Becomes Fable 5.1

Anthropic shipped a new flagship model.

Claude's Top Tier Becomes Fable 5.1

Tokens and the Context Window: The Root Concepts of Billing and Limits
Compared to before

Until last month, Opus 5 was effectively the top model, with claude-opus-5 doing the heavy lifting.

What changed

Anthropic released Fable 5.1, a new flagship above Opus 5. Pricing is higher than Opus 5, but the context window is unchanged from the old Fable 5.

Why it matters

More headroom for top-tier work, but at roughly double the price — if Opus 5 already covers your needs, this is just a bigger bill.

B · Theme of the day

Coding Assistants Stop Betting on One Model

GitHub Copilot starts switching brains per task.

Copilot Gets Smarter, Cheaper Multi-Model Runs

GitHub CopilotGitHub Copilot
Compared to before

Until last month you picked one model in the picker and stuck with it for the whole session.

What changed

Copilot CLI's HydraFusion auto-picks among multiple models per task; on TerminalBench 2.1 it scores higher than Claude Opus 5 alone, at lower cost.

Why it matters

Less manual model-swapping for mixed tasks — though for a single quick fix, the gain is marginal.

C · Theme of the day

Specialist AI Spreads Into Robots and 3D

Purpose-built models keep expanding beyond chat.

Tripo's Fast Model Gets Game-Ready Meshes

Tripo AITripo AI
Compared to before

Until early this month, Smart Mesh was the cheap, fast option — game-ready topology meant switching to the pricier H3.1 flagship.

What changed

Smart Mesh P2.0 Preview shipped Aug 28: native quad topology up to 25,000 quads (~50K triangles), selectable in Tripo Studio.

Why it matters

The budget tier can now deliver game-ready meshes on its own, cutting reliance on H3.1 — though intricate organic models still favor H3.1.

Mistral Enters Robotics With Robostral Navigate

MistralMistral
Compared to before

Until now, Mistral's specialist models covered documents, code, speech, and safety — nothing built for robots.

What changed

Mistral shipped its first robotics model, Robostral Navigate, built for embodied navigation — joining its OCR, code, speech, and safety specialists.

Why it matters

A new option if you build robotics; if you just chat or code, this announcement doesn't touch you.

Today's Tips

2

Practical ways to use these tools, picked up from social media and community threads.

T1 · Theme of the day

Assistants Learn Your Everyday Details

Three hidden tricks in chat-style AI.

A Notion "Custom Skill" Is Just a Page

Notion AINotion AI
How to do it

Convert any Notion page into a reusable skill straight from its ••• menu — no separate skill-builder needed.

Why it helps

You can start automating repeat work using the page habits you already have, no new tool to learn.

Tell Gemini Where You Left Something

GeminiGemini
How to do it

Say "my passport is in the bedroom drawer" and Gemini saves it to a new Remembered tab you can ask about later.

Why it helps

Handy as a substitute for physical trackers — works even on items without a tracker tag.

Gemini Describes What's In Front of You

GeminiGemini
How to do it

Share your camera in Gemini Live to get Guided Vision — it describes your surroundings out loud, handy for fine print or a dim menu.

Why it helps

Useful when you can't easily read something yourself — small text, low light, unfamiliar signage.

T2 · Theme of the day

Shave a Step Off the Creative Tools

Fewer clicks in image and editor AI.

Preview Midjourney's Next Version Early

MidjourneyMidjourney
How to do it

Add --preview to the end of a prompt to try the look of the next version (e.g. V8.2) before general release.

Why it helps

Lets you get a feel for the upcoming style shift before it ships. Skip it if you're happy with the current look.

Windsurf Adds a Free Lightweight Model

WindsurfWindsurf
How to do it

Cascade Base is a free, fast model built for small, low-complexity edits — switch to it when deep reasoning isn't needed.

Why it helps

Skips the overhead of a heavy model for trivial fixes, saving both wait time and credits.

Turn Repeat Tasks Into Windsurf Slash Commands

WindsurfWindsurf
How to do it

Save a repetitive task as a custom Cascade Workflow and invoke it later as a slash command instead of retyping instructions.

Why it helps

Cuts the busywork of rewriting the same instructions every time you run a routine task.

Archive

Past updates

A daily archive of changes actually applied to the site.