The weekly loadout for AI builders. What changed, what to try in ~10 min, what to skip. Free newsletter: https://t.co/APXwARzVXspatchnotesai.comJoined October 2026
[NEW] Google Play: build an Android app in Google AI Studio and run it on your own device with a standard Google account. No developer account; the app stays private. Rolling out to select markets incl. the U.S., 18+. Friends-and-family sharing 'in the coming months', per Google.
Sonnet 4.5 retires Nov 30 on the Claude API (Bedrock and Google Cloud set their own dates). Sonnet 5.5 lists $2/$10 per 1M tokens vs $3/$15. Anthropic says the newer tokenizer gives ~30% more tokens for the same text, so our own estimate is ~13% cheaper (before caching), not 33%.
[NERF] OpenAI is retiring its Evals platform. Existing evals become read-only on Oct 31, and the Evals dashboard and API are scheduled to shut down Nov 30. OpenAI's docs point to Promptfoo as a migration path. Export what you need before then.
[NERF] OpenAI shuts down a batch of older API models on Oct 23, three days out. The list includes gpt-4-turbo, gpt-image-1, o1-2024-12-17, o3-mini-2025-01-31 and o4-mini-2025-04-16. Search your code for those names. OpenAI lists a replacement for each.
Where Haiku 5.5 is a good fit: subagents, summaries and classification, up to 100K-token prompts at the quoted price. Test on a copy of your workload first.
The week's other changes, and what to skip, are in Issue #1 of Patch Notes AI. Free, weekly: patchnotesai.com
Prefill. Haiku 5.5 rejects an assistant-message prefill with a 400 error, even with thinking off (Haiku 4.5 accepted one). Any trick that starts the answer for the model (forcing JSON, for example) needs a different approach.
Haiku 5.5 is 10x cheaper than Haiku 4.5 on paper: $0.10/$0.50 per 1M tokens vs $1/$5 (for prompts up to 100K tokens).
Before you swap the model ID, here are 3 things that can eat the savings or break your app. A short thread:
Claude Haiku 5.5 is 10x cheaper than Haiku 4.5 on paper (prompts up to 100K tokens). Three things can eat the savings: a new tokenizer, sampling params, prefill. Our own estimate: closer to 7-8x. About a minute:
Skip list, with receipts:
- Launch-day benchmarks: Argon and Muse Wonder both lead with their own numbers.
- A blind Haiku 4.5 to 5.5 swap: ~30% more tokens for the same text, and temperature/top_p/prefill can now error.
- Budgeting Argon at $2/$10: it's $4/$20 after Nov 30.
Gemini 4 Argon is out. Where you can use it:
- Paid Gemini API
- Gemini app: Pro and Ultra
- Antigravity: Ultra only
- GitHub Copilot: Pro+, Max, Business, Enterprise (gradual rollout)
API intro price: $2/$10 per 1M tokens. Ends Nov 30, then $4/$20.
2.7M Followers 47 FollowingThe official handle for NVIDIA. Blog: https://t.co/JAn5eKOTBT Support: https://t.co/6ln5FVnA2o All our social media: https://t.co/Uc56dL57Dh
599K Followers 53 FollowingThe Gemini app turns research into reality, bringing frontier AI experiences like Omni, Nano Banana, and more to over a billion people.
1.9M Followers 2 FollowingClaude is an AI assistant built by @anthropicai to be safe, accurate, and secure. Talk to Claude on https://t.co/ZhTwG8d1e5 or download the app.
8.7M Followers 11 FollowingSteam, The Ultimate Online Game Platform. Follow us here for news on the latest releases and special promotions!
For support: https://t.co/ecoSK7uTBx
1.9M Followers 1K FollowingCo-Founder of Coursera; Stanford CS adjunct faculty. Former head of Baidu AI Group/Google Brain. #ai #machinelearning, #deeplearning #MOOCs
601K Followers 589 FollowingFounder of the world’s most read daily AI newsletter @therundownai. Sharing the latest developments in the world of artificial intelligence.
1.6M Followers 268 FollowingThe engine room of @Google. Building AI safely and responsibly to solve the world’s most complex problems. Join us: https://t.co/jUHQA27iBL
1.8M Followers 2 FollowingWe're an AI safety and research company that builds reliable, interpretable, and steerable AI systems. Talk to our AI assistant @claudeai on https://t.co/FhDI3KQh0n.
5.5M Followers 4 FollowingOpenAI’s mission is to ensure that artificial general intelligence benefits all of humanity. We’re hiring: https://t.co/dJGr6LgzPA