Terminal X Uses GPT-6.1 Sol for Investment Workflows

Terminal X uses OpenAI models including GPT-6.1 Sol and GPT-6 Luna inside agents for its Capital Memory operating system. Across 500 scored judgments, Terminal X's Eval Agent running on GPT-6.1 Sol agreed with a senior human reviewer 88.6% of the time.

According to Terminal X, this represents an increase of 220 basis points over the next-best model, which reached 86.4% agreement. Terminal X also stated that GPT-6.1 Sol caught about 25% more departures from a fund's mandate than the next-best model, such as recommendations for distressed credit positions instead of take-private candidates.

To deploy these agents, Terminal X says its engineers—former finance and investment professionals at JPMorgan and Blackstone—document firm processes and encode them into customized agents deployed on-premise or in the firm's cloud. The agents process documents in English, Korean, Japanese, and Chinese. Workflow stages requiring judgment use GPT-6.1 Sol, while high-volume steps use GPT-6 Luna.

An Eval Agent scores outputs against a 54-criterion rubric covering source selection, narrative consistency, and alignment with house views. Each failed check provides a reason and corrected text to update a firm Memory record. Terminal X reported that GPT-6.1 Sol ran the workflows at approximately 30% lower cost per query than the next-best model at similar speed. The company stated that one private equity customer achieved $2.4 million in annualized savings from automated due diligence and investment memo workflows. Terminal X says it has built 1,700 firm Memories, run over 1 million custom workflows, and plans to extend firm Memories from desks to individuals.