Samaya AI has released FrontierFinance, a new public benchmark for evaluating AI agents in investment workflows. The benchmark comprises 220 diverse queries and 11,543 expert-crafted rubrics, covering the full investment process from idea screening to portfolio tracking. Samaya's proprietary AI system outperformed all tested models, including Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol. Samaya's high-effort configuration achieved 56% accuracy at half the cost of Fable 5, whilst its low-effort version scored 50.8% at a quarter of the cost. Claude Fable 5 was the best-performing frontier model at 49.2%, followed by GPT-5.6 Sol at 46.8%. The benchmark evaluates AI capabilities across investment stages including research, financial modelling, and catalyst monitoring. Samaya AI builds expert AI agents for investment professionals and serves tens of thousands of investors globally.

PitchBook has announced a premium partnership with Samaya AI, integrating its private capital market data into Samaya's AI agent platform. The collaboration enables investment professionals to access private market intelligence through Q&A prompts and agent workflows without switching platforms. The integration combines PitchBook's data on private companies, investors, deals and funds with Samaya AI's large language models. Users can surface company profiles, run deal comparisons, prepare for meetings and source transaction multiples with full traceability to primary sources. The partnership is part of PitchBook's expanding AI collaborations, which include Anthropic, Hebbia, OpenAI and Perplexity. PitchBook, a Morningstar subsidiary founded in 2007, serves over 100,000 clients globally with private capital market intelligence.