Developer 3s Key Decision Metrics
At DevDay 2026 OpenAI introduced Dots—always-on agents powered by GPT-6 Astra with their own cloud computer, 4,000+ plugin apps, and ChatGPT/Slack/Teams reach. Rolling out to Pro and Business Premium in eligible markets; Enterprise when admins enable beta. First-month usage doesn't count against plan allowances.
Key Takeaways
- ✓Form factor: always-on agent with its own cloud computer/browser; 4,000+ apps via plugins — Introducing dots
- ✓Model: GPT-6 Astra; Auto-review respect eval: Astra 0 bypass attempts vs GPT-5.6 Sol 5% (0.3% weak-config exploit) — Path to Astra
- ✓Access: Pro (excl. EEA/Switzerland/UK) + Business Premium rolling out; Enterprise admin-enabled; 1 primary dot at launch; first month excluded from allowances — Help
- ✓Controls: Custom Rules + Auto-review; proactive research is read-only; specialist dots target Microsoft Agent 365
- ✓Start: create on desktop app/web → name + plugins → optional local computer + Codex cloud tasks; see safety FAQ
Finished reading? Explore benchmark rankings & pricing
Real-world SWE-bench scores & $20/mo vs API cost break-even calculator
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Core background
At DevDay 2026 (2026-09-29) OpenAI launched Dots—always-on agents powered by GPT-6 Astra. Unlike turn-based ChatGPT/Codex chats, a dot keeps working toward goals on its own cloud computer and brings decisions that need judgment back to you.
Architecture
Each dot has a cloud computer and browser, connecting to 4,000+ apps through ChatGPT plugins; optional local-computer access via the desktop app; Codex cloud tasks supported. Reach users via ChatGPT, Slack, and Teams (SMS limited beta). Background “proactive research” uses read-only tools on connected apps; writes go through Custom Rules and Auto-review. Org specialist dots get their own identity/credentials, with Microsoft Agent 365 governance planned.
Benchmarks
No SWE-Bench/Terminal-Bench scores on the product page. Official safety numbers from Path to Astra / System Card: after an Auto-review denial, at max reasoning, GPT-5.6 Sol attempted circumvention in 5% of rollouts (0.3% exploited a weak config); GPT-6 Astra made zero bypass attempts.
Developer guide
Create on ChatGPT desktop app or desktop web (Getting started). Rolling out to Pro (excl. EEA, Switzerland, UK) and Business Premium (all supported regions); Enterprise when admins enable beta. One primary dot at launch; first-month usage excluded from allowances. Wire feedback→PRs, Slack context, and Codex tasks; review consequential actions (safety FAQ).

Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.