$\texttt{YC-Bench}$: Benchmarking AI Agents for Long-Term Planning and Consistent Execution
This addresses the need for better benchmarks to assess long-term planning and consistent execution in AI agents, which is crucial for deploying them in complex real-world tasks, though it is incremental as it builds on existing benchmarking efforts.
The paper tackles the problem of evaluating AI agents' ability to maintain strategic coherence over long horizons by introducing YC-Bench, a benchmark simulating a startup over a one-year horizon, and finds that only three out of twelve models consistently surpass the starting capital, with Claude Opus 4.6 achieving the highest average final funds at $1.27M.
As LLM agents tackle increasingly complex tasks, a critical question is whether they can maintain strategic coherence over long horizons: planning under uncertainty, learning from delayed feedback, and adapting when early mistakes compound. We introduce $\texttt{YC-Bench}$, a benchmark that evaluates these capabilities by tasking an agent with running a simulated startup over a one-year horizon spanning hundreds of turns. The agent must manage employees, select task contracts, and maintain profitability in a partially observable environment where adversarial clients and growing payroll create compounding consequences for poor decisions. We evaluate 12 models, both proprietary and open source, across 3 seeds each. Only three models consistently surpass the starting capital of \$200K, with Claude Opus 4.6 achieving the highest average final funds at \$1.27 M, followed by GLM-5 at \$1.21 M at 11$\times$ lower inference cost. Scratchpad usage, the sole mechanism for persisting information across context truncation, is the strongest predictor of success, and adversarial client detection is the primary failure mode, accounting for $47\%$ of bankruptcies. Our analysis reveals that frontier models still fail through distinct failure modes such as over-parallelization, demonstrating the capability gaps for long-horizon performance. $\texttt{YC-Bench}$ is open-source, reproducible, and configurable.