Tech • AI • Robotics • Game
AI News: ChatGPT Dots, Major Grok Bot Upgrades, GPT Sol 6.1 (+ more)
9.4/10TL;DR
OpenAI, Anthropic, and xAI rolled out major AI products and model updates, centered on proactive assistants, collaborative workspaces, lower-cost frontier models, and team-based agents.
KEY POINTS
OpenAI launches DOTS assistant
OpenAI introduced DOTS, a personal AI assistant designed to act proactively rather than wait for prompts. It can be reached through the ChatGPT app, phone calls, SMS, email, and Slack, and can monitor connected apps to surface urgent items or complete delegated work. The product combines Codex, ChatGPT Work, a cloud browser, and connected apps so tasks can run off-device in the cloud.
Permissions and pricing for DOTS
The assistant includes permission controls for app access and approvals before taking sensitive actions. A first DOT is included at no extra charge for Pro and Business Premium users, and simple conversation with the assistant does not count against standard ChatGPT usage limits. Usage is charged only when the assistant starts or manages tasks through tools such as Codex or ChatGPT Work.
ChatGPT Space targets AI-native teamwork
OpenAI also released ChatGPT Space, a shared workspace that resembles a hybrid of a document tool and an AI collaboration hub. Teams can organize work into spaces, pages, and subpages, while AI assistants can edit pages, generate assets, and update dashboards directly inside the workspace. The product is available for Pro, Business, and Enterprise users across web and desktop, with mobile rollout expected.
Plugins expand workspace automation
A key feature of ChatGPT Space is plugin support, letting third-party apps operate inside the workspace. In one example, a creative app generated multiple product ad images and videos for Instagram and inserted them back into a page with format notes. The setup points to OpenAI’s strategy of turning ChatGPT into a unified operating layer for documents, media creation, and app actions.
GPT 6.1 Soul cuts model costs sharply
OpenAI released GPT 6.1 Soul, positioned as near-Astra-level performance at a much lower price. Pricing was set at $2 per million input tokens versus $10 for GPT 6 Astra, $0.10 for cached inputs versus $1, and $10 per million output tokens versus $50. The model is available in ChatGPT Work, Codex, and the API for Plus, Pro, Business, and Enterprise users.
New $500 plan raises usage ceiling
The company added a $500-per-month subscription tier that includes the new Ultraast model, removes the previous 5-hour cap, works across partner apps, and offers 25 times the usage of the Plus plan. The package appears aimed at heavy users running frequent AI workflows across multiple tools.
Decisions API speeds up computer-use agents
Another release, the Decisions API, is intended to improve the speed of AI agents making interface-level decisions. That matters for so-called computer-use tasks such as navigating websites or booking travel, where previous systems often lagged between clicks. Faster decision loops could make practical agents more usable for real work.
Anthropic updates Claude Sonnet 5.5
Anthropic introduced Claude Sonnet 5.5 days after launching Opus 5.5. The company said Sonnet 5.5 is a clear upgrade over Sonnet 5, runs 30% faster, and costs up to 30% less for many tasks. It also signaled that Haiku 5.5, the lower-cost member of the family, is likely to follow soon.
Sonnet 5.5 emphasizes cost-efficient coding
Anthropic is positioning the new model as a strong coding option with clearer communication and better collaboration. In cited benchmarks, Sonnet 5.5 reached 70.6% on agentic coding compared with 10.3% for Sonnet 5, and even outscored Opus 5.5 on that metric at 66.4%. The company and early testers described it as delivering high-end performance at far lower cost per task.
xAI adds Grokbot Teams and a primary bot
xAI released Grokbot Teams, a shared bot system for workplaces. Team members can use one shared bot with common files, instructions, and skills, while their individual conversations remain private and keep separate memory and context. A new primary bot feature was also added to act as a central assistant that can route work to more specialized bots, effectively creating a chief-of-staff layer for users managing multiple AI agents.
CONCLUSION
The week’s releases show the AI industry shifting from standalone chatbots toward integrated assistants that can act, collaborate, and work across software tools. The main competitive front is no longer only raw model quality, but how much useful work can be delivered at lower cost and with less friction.
Ask a question