Microsoft’s new Copilot moves toward persistent agents
Home, Code and Autopilot expand Copilot beyond a chat interface, with staged access and controls for agent spending.
News brief · 1 sourceRead →
AI / NEWS BRIEF
GPT-6.1 Sol puts the cost of capable agents in focus
OpenAI’s latest Sol release targets coding, computer use and professional work at lower prices than Astra. The performance claims still need independent testing.
What changed
OpenAI has introduced GPT-6.1 Sol as an upgrade to GPT-6 Sol. The company reports improvements in software engineering, computer use and professional document workflows. Its standard API pricing is $2 per million input tokens and $10 per million output tokens, with cached input priced at $0.10 per million tokens.
Why it matters
OUR ANALYSISFor teams running repeated agent workflows, the interesting question is how much useful work each dollar buys. A lower-priced model that completes a demanding assignment reliably could make automation practical for more workloads. That is the opportunity described by this launch, rather than proof that every task should move to Sol.
The important limitation
OUR ANALYSISThese comparisons come from OpenAI’s announcement and its described evaluation settings. They are provider-reported results, not AIQuantumScore research evaluations. REVAL testing must measure the exact model and configuration separately before a proprietary research score can be published.
More to follow
Duke’s 13-ion simulator probes how particle pairs form
A trapped-ion experiment recreates string-breaking dynamics, offering a controlled way to investigate high-energy physics without claiming a general quantum-computing advantage.
News brief · 1 sourceRead →