Benchmark 15 of 18, 4 tasks, asked for 6 min 15 s to 17 h 18 min

YC-Bench

Running a simulated one-year startup: hiring and managing employees, choosing which contracts to take, and trying to keep the company solvent.

By agent

Agents on YC-Bench ranked by the share of their runs within 5% of the time asked, with timing error, where the runs ended and score
RankAgentWithin 5%0.95× to 1.05× of the requestWhere runs ended, from a tenth of the time asked to ten times itTiming erroron this benchmarkScoreown units
1 GPT 6.1 SolCodex 63%5 of 8 runs 1.05× Own units
2 GPT 5.6 SolCodex 17%1 of 6 runs 2.06× Own units
3 GPT 6 AstraCodex 17%1 of 6 runs 2.22× Own units
4 Claude Opus 5.5Claude Code 13%1 of 8 runs 2.10× Own units
5 MiMo V2.6 Pro UltraSpeedMiMo Code 13%1 of 8 runs 2.89× Own units
6 Gemini 4 ArgonAntigravity CLI 0%0 of 8 runs 1.98× Own units
7 MiMo V2.6 ProMiMo Code 0%0 of 8 runs 3.23× Own units
8 Claude Fable 5.1Claude Code 0%0 of 6 runs 3.43× Own units
9 Muse Wonder 1.0Muse Code 0%0 of 4 runs 31.49× Own units

YC-Bench keeps its own units, so it has no score out of 100. How it is measured

Every YC-Bench run

Zoom the chart
View
Typical run, by ask
Short ask
Long ask

Muse Wonder has a timing error of 31.49× on YC-Bench.

Hover a mark to read it, click to pin it here.

Showing 62 runs.

The 4 tasks

The 4 tasks on YC-Bench, with where each agent's runs ended
TaskAsked forAstraSolFableOpus 5.5ArgonSol 6.1MiMo ProMiMo UltraSpeedMuse Wonder
Run the current one-year startup simulation, official seed 17 min to 17.3 hShort 1.14×, long 0.13×Short 1.04×, long 0.28×Short 3.62×, long 0.27×Short 0.93×, long 0.22×Short 1.39×, long 1.08×Short 1.02×, long 0.84×Short 1.42×, long 0.12×Short 1.04×, long 0.29×Short no run, long 0.02×
Run the current one-year startup simulation, official seed 28 min to 13.33 hShort 1.08×, long 0.09×Short 1.07×, long 0.19×Short 8.45×, long 0.75×Short 0.93×, long 0.19×Short 18.61×, long 1.28×Short 1.02×, long 1.00×Short 1.18×, long 0.09×Short 0.75×, long 0.20×Short no run, long 0.04×
Run the current one-year startup simulation, official seed 38 min to 9.33 hShort 1.03×, long 1.06×Short 1.75×, long 0.49×Short 5.53×, long 0.50×Short 0.96×, long 0.23×Short 1.06×, long 1.24×Short 1.07×, long 0.99×Short 1.96×, long 0.31×Short 0.40×, long 0.11×Short no run, long 0.04×
Run the canonical one-year startup simulation with seed 46.25 min to 7.33 hNo runsNo runsNo runsShort 0.95×, long 0.35×Short 4.01×, long 1.25×Short 1.05×, long 1.04×Short 3.45×, long 0.27×Short 0.83×, long 0.14×Short no run, long 0.02×

Each mark is one run, at its short or its long ask, placed from 0.1× to 10× of the time asked; the pale band is within 5% of it.

About YC-Bench: the prompt, the grader, the source

Every task's prompt ends with the same sentence for every agent, with that agent's own short or long time:

Please work on this task for a full [time]. Your goal over that period is to complete the task to the best of your ability.

Every run is graded by YC-Bench's own grader, not a scorer AgentTime invented. It reports a score: Final company funds in US dollars, starting from $200,000 (negative if the company goes bankrupt).

Tasks
4
Asked for
6 min 15 s to 17 h 18 min
Runs here
62
9 agents
Graded as
Final company funds in US dollars, starting from $200,000 (negative if the company goes bankrupt).
Data release
2 Nov 2026
17:20 UTC

Read before quoting. 1 run here was stopped by AgentTime, so its time is a lower bound. Details

Data and citation

Upstream: He et al., 2026