September 28, 2026
Basis completes a tax workbook 2x faster with GPTβ6 Astra
GPTβ6 Astra took 50% less time to finish complex tasks vs. GPTβ5.6 Sol and showed a deeper understanding of user intent.
Workbook test
50%
Less time to complete a 50-tab tax workbook vs. Sol in Basisβs test.
Internal evaluations
20%
Approximate improvement in Basisβs internal evaluation scores.
Basisβ (opens in a new window) builds AI agents to automate much of the manual work that accountants do each day, helping them shift their time from repetitive tasks to strategic work. The companyβs research focuses on agents that can reliably complete long tasks, and with GPTβ6 Astra, itβs seeing a stronger understanding of what accountants want to accomplish.
βGPT-6 Astra does a better job of really understanding the intent of the user and the problem.β
Completing a 50-tab tax workbook in half the time
Basis compared GPTβ6 Astra and GPTβ5.6 Sol on a complicated tax workbook with 50 tabs. The task was to complete the workbook accurately and reliably, and GPTβ6 Astra was markedly faster.
βGPT-6 Astra is able to complete that workbook in half the time that GPT-5.6 Sol is able to.β
Basis also noted that GPTβ6 Astra makes better decisions at the start of a task, helping Basisβs agents take a more direct path through the work with less time spent correcting mistakes. Troyanovsky says that also makes the model more efficient in its use of tokens.
Matching reasoning effort to the task
Basis also has GPTβ6 Astra adjust how much reasoning it uses as a task progresses, dialing up computation when a step is difficult, and using less when a step is easier. The model can make these adjustments while keeping its cache intact. Troyanovsky says this helps reduce cost and response time, making long-running tasks more economical for Basis and its customers.
Building confidence in real-world use
Basis saw about a 20% improvement in its internal evaluation scores with GPTβ6 Astra, driven by better understanding of user intent, including when to ask questions, flag assumptions, and follow instructions.
Basis evaluates how its agents work and their final answers, including whether they follow templates, consult primary sources for tax questions, and check their own work. GPTβ6 Astra can infer these expectations from a broader context, with fewer explicit instructions.
That reduces the need for Basis to write rules for individual situations and gives the team more confidence that its agents can handle situations beyond those covered in internal tests.

