“I don’t use Astra every day”: OpenAI developer lead tells TNS why

An OpenAI developer lead shared insights into the recent wave of AI releases, noting, "I don’t use Astra every day."

During the 2026 DevDay, OpenAI unveiled numerous new models, tools, and agents, sparking discussions around operational hurdles such as cost, speed, and maintaining team alignment.

Vaibhav “VB” Srivastav, the developer experience and community lead for EMEA at OpenAI, discussed these features with The New Stack ahead of the November 3 Dev Day Exchange in London. He addressed how developers navigate this rapid influx of capabilities.

Srivastav highlighted GPT-6.1 Sol as a key development. It achieves roughly 95% of Astra’s performance while costing one-fifth as much. He stated, "I don't use Astra every day. I do use GPT-6.1 Sol as my main go-to."

When asked about the performance gap, Srivastav explained that evaluations like Terminal-Bench 4.0 and SWE-bench show GPT-6.1 Sol is empirically very close to GPT-6 Astra. Furthermore, internal testing showed a strong adoption trend toward Sol.

For deciding between the models, Srivastav advised using Astra for novel or unfamiliar problems to find potential blind spots. He noted, "If I use GPT-6.1 Sol to work on that feature and then use the same model for code review, it’s like checking my own work. I’d be biased toward saying it’s good."

Ultrafast mode is reserved for critical, real-time scenarios like fraud detection. This mode allows both GPT-6 Astra and GPT-6.1 Sol to deliver outputs exceeding 300 tokens per second, though at a higher price.

Regarding governance, Srivastav mentioned that enterprise features are off by default. He championed Auto-review, a system where a lightweight model vets actions against a defined policy before the main model proceeds. This feature was recently made free for Codex and ChatGPT subscribers.

A major announcement was the introduction of OpenAI dots—persistent, always-on agents. These agents scan connected tools, like Slack and GitHub, to surface items requiring attention or to execute tasks autonomously, such as merging approved pull requests.

Srivastav shared how his own dot monitors his workflow across various platforms. He explained that this automation frees up time previously spent sifting through notifications, allowing him to focus on higher-level tasks.