AI Dev Weekly #28: GPT-6.1 Sol, Claude Sonnet 5.5, Dots and NVIDIA OpenShell
AI Dev Weekly is a Thursday series where I cover the week's most important AI developer news, with my take as someone who actually uses these tools daily. Last week brought a crowded model launch cycle. This week turned those models into systems. OpenAI replaced the two-week-old Sol baseline with GPT-6.1 Sol, added native delegation and browser execution to its agent stack, and introduced an…
Last week's AI Dev Weekly highlighted the launch of GPT-6.1 Sol by OpenAI, Anthropic's Claude Sonnet 5.5, NVIDIA's OpenShell, and the introduction of ChatGPT's always-on Dots agents. This week, those new models have transformed from mere releases into fully operational systems.
OpenAI's GPT-6.1 Sol, launched on September 29, builds upon its predecessor with a 1,050,000-token context window, 128,000 maximum output tokens, and support for multi-agent delegation in beta. Priced at $2 per million input tokens and $0.10 for cached input, it offers near-Astra performance for a fraction of the cost. The model's multi-agent support allows for more complex agent designs, but it also necessitates careful management of permissions and traceability.
Anthropic's Claude Sonnet 5.5, released on the same day, aims to bring Opus-like performance at Sonnet pricing. With a 30% faster response time and the potential for 30% lower costs per task, it's positioned as a strong default candidate for routine coding and professional workflows. However, its integration contracts differ from previous versions, requiring developers to account for changes in tool selection, thinking modes, and output formatting.
NVIDIA's OpenShell focuses on agent security and application-layer protection. By securing agents below the prompt and application layers, it provides a more robust foundation for AI system development. The details of OpenShell are not explicitly mentioned in the source, but its emphasis on security is a crucial aspect of the current AI development landscape.
Finally, OpenAI's Dots agents, introduced during DevDay, offer always-on functionality in ChatGPT. These agents can operate independently between conversations, perform deeper work through delegation, and are tailored for ongoing responsibilities. Currently limited in availability, Dots represent a shift towards more persistent AI applications.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
