Karina Nguyen demonstrates Inkling 2 Small independently parsing a PSEG 10-Q in a financial diligence task and discovering a $1.2 billion collateral release
One of my favorite traces: when the filing tool returned only PSEG’s table of contents, Inkling 2 Small agent independently downloaded th...
AI BriefingKarina Nguyen shared a real-world trace of an Inkling 2 Small agent: when the filing tool returned only PSEG's index, the agent independently downloaded the complete 3.7MB 10-Q, parsed it with Python, and discovered a $1.2 billion collateral release that needed to be excluded from operating cash flow. It scored 0.82, higher than the next-best model's 0.67. The cited @thoughtfullab post said it performed well on DiligenceBench. The model is an MoE with 594B parameters and 17B active parameters, priced at $0.30/M input tokens and $1/M output tokens.
Source: Karina Nguyen · x.com
Event follow-up1
View all event coverage- Thinking Machines Lab releases Inkling 2 Small: a 594B-parameter MoE model, with Apache 2.0 weights to be released next weekThinking Machines Lab:News · 11 days ago