Thrive Holdings Launches Real-Work Benchmark Across 75 Companies

Thrive Holdings released a new benchmark evaluating AI models on real workflows from 75+ accounting and IT companies, measuring accuracy, time, and cost. In tax filing, Opus 5.5 led while GPT-6.1 Sol trailed by just 2.1 points at a quarter of the cost, with model rankings flipping depending on the task.

2026-10-16 ~ 2026-10-16 · 4 related posts

• Thrive Holdings benchmarks AI on real tax and IT work across 75+ firms — BorisMPower · 2026-10-16
• Thrive Holdings launches real-work AI evals from 75+ businesses, benchmarking 13 models — samaysham · 2026-10-16
• Thrive Holdings: measuring model ROI with evals on real daily workflows — samaysham · 2026-10-16
• Holdings evals: Opus 5.5 leads tax prep while GPT-6.1 Sol matches within 2.1 pts at 1/4 the cost — samaysham · 2026-10-16