The agent is being graded on endurance
the agent is being graded on endurance: yc-bench runs models as startup ceos for a simulated year and only three of twelve kept the company above starting capital, while agentworld pushes 3 to 20 agents through 50-round collaborations in a game sandbox. gpt-5.6 sol leads a 55-field professional workflow eval by 13 points.
the long game is where the leaderboards stop lying.
Context
YC-Bench, from Collinear AI and an arXiv paper, is a long-horizon deterministic benchmark in which an LLM agent runs a simulated AI startup over one simulated year, managing resources and contracts. It tests strategic coherence under delayed feedback and compounding early mistakes.
AgentTime also runs YC-Bench, grading timing error and where runs ended.
YC-Bench as a simulated one-year startup benchmark matches the sources. Only three of twelve models keeping the company above starting capital, AgentWorld's 3 to 20 agents in 50-round game collaborations, and GPT-5.6 Sol leading a 55-field professional workflow eval by 13 points were not seen in the sources read, so unsupported here, not refuted.
Related work
- GitHub: collinear-ai/yc-bench ↗Benchmark repo.
- arXiv: YC-Bench ↗Paper.
- AgentTime: YC-Bench ↗Timing results.
Watch next
- Find the YC-Bench leaderboard with per-model net worth.
Sources
- GitHub, collinear-ai/yc-benchgithub.com
- arXiv 2604.01212arxiv.org
- YC-Bench project pagecollinear-ai.github.io
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 2 October 2026 at 16:41 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →