Oct 20, 2026, 7:00 AMArtificial Intelligence
Terminal X Eval Agent Matches Investor Judgment 88.6% With GPT-6.1 Sol
Terminal X says its Eval Agent matches senior investor judgment 88.6% and catches 25% more mandate deviations using GPT-6.1 Sol across 54 evaluation criteria.
Listen to this briefingAudio briefing
Summary
On October 20, 2026, Terminal X said its Eval Agent matched senior investor judgment at an 88.6% rate using GPT-6.1. The agent grades outputs against a 54-criterion rubric and, with GPT-6.1 Sol, caught 25% more deviations from a fund’s mandate.
Details remain limited. The available summary does not specify the evaluation sample, methodology, comparison baseline, participating funds, deployment plans or pricing, restricting assessment of the claimed gains.
Positives
- Terminal X reports an 88.6% match between Eval Agent results and senior investor judgment.
- A 54-criterion rubric gives the Eval Agent a structured framework for grading outputs.
- GPT-6.1 Sol helped detect 25% more deviations from a fund’s mandate.
Risks & concerns
- The available summary does not identify the baseline behind the claimed 25% improvement.
- No sample size, testing methodology or participating funds are specified.
- Deployment plans, availability and pricing for Terminal X’s Eval Agent remain undisclosed.
Primary sourceOpenAI Newshttps://openai.com/index/terminal-x
Read full article