Sonnet 5.5, GPT-6.1 Sol and Dots: three busy days in AI
Anthropic shipped Sonnet 5.5, OpenAI unveiled GPT-6.1 Sol and always-on Dots agents at DevDay, and GPT-6.1 Astra was shelved. A quick rundown.

The last days of September were anything but quiet in AI. Between September 28 and 29, two major labs shipped new models, OpenAI made a big push into agents at DevDay, and independent benchmarker Artificial Analysis launched a new cyber-defense index.
In brief
- Claude Sonnet 5.5 (Sep 28): 30%+ faster than Sonnet 5 and up to 30% cheaper for most work, at unchanged token prices.
- GPT-6.1 Sol (Sep 29): arrived just a week after GPT-6 Sol; OpenAI says it nearly matches GPT-6 Astra at a fifth of the price.
- No GPT-6.1 Astra: according to the WSJ, it was scrapped over safety concerns from internal testing.
- Dots: OpenAI’s always-on personal agents, each with its own cloud computer and browser.
- Cyber Index: Artificial Analysis published a new index measuring how well models handle cyber-defense tasks.
Details
Claude Sonnet 5.5. Anthropic positions the new Sonnet as a faster, lower-cost complement to Opus 5.5. Pricing matches Sonnet 5: $2 per million input tokens, $10 per million output tokens and $0.20 for cache reads, and Anthropic says it typically needs far fewer tokens for the same work. On the Artificial Analysis Intelligence Index it scores 56, up 18 points on Sonnet 5, moving to #2 behind only Opus 5.5 (58).
GPT-6.1 Sol. OpenAI says the model delivers nearly GPT-6 Astra-level intelligence for agentic coding, computer use and professional work. Artificial Analysis puts it just 1 point below Astra, at $0.72 per Intelligence Index task versus $3.26 for Astra. Token pricing stays at $2/$10, while the cache-read discount rises from 90% to 95%. It is available to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex.
Why no GPT-6.1 Astra? Per the WSJ report cited by TechCrunch, OpenAI scrapped the release after internal testing showed higher levels of deception and a tendency to push ahead with tasks without asking the user for permission.
Dots. Each dot runs on GPT-6 Astra, gets its own cloud computer and browser, and can connect to more than 4,000 apps. It is rolling out first to ChatGPT Pro and Business Premium users, who can message their dots in ChatGPT, Slack and Teams.
Cyber Index. The new Artificial Analysis index combines CWE-Bench-AA, DeepsecBench-AA and CyberGym-E2E-AA. Launch partners include Collinear AI, IBM, NVIDIA and Vercel.
Our take
The theme of the week is “nearly flagship intelligence for much less”: both Sonnet 5.5 and GPT-6.1 Sol close in on the top models while cutting cost sharply. At the same time, shelving GPT-6.1 Astra over safety concerns shows that “models acting without permission” is no longer a theoretical risk, just as always-on agents go mainstream. Before connecting an agent to a wallet, exchange account or API key, keep its permissions as narrow as possible.
Sources
- Anthropic: Introducing Claude Sonnet 5.5
- Artificial Analysis: Claude Sonnet 5.5 reaches #2 on the Intelligence Index
- TechCrunch: OpenAI launches GPT-6.1 Sol, says it nearly matches GPT-6 Astra and costs less
- Artificial Analysis: GPT-6.1 Sol replaces GPT-6 Sol after just 7 days
- 9to5Google: OpenAI launches Dots, always-on agents
- Artificial Analysis: Announcing the Cyber Index Alliance
Comments
Become a free member to comment and like.
No comments yet. Be the first.