Skip to content
Agent
VFS
Spec
01
VFS.md
Benchmark
VFS-BENCH 1.0
01
Leaderboard
02
Tasks
ALL BENCHMARKS
Try in Browser
Blogs
Contribution
Gym
People
SPEC
VFS.md
BENCHMARK
VFS-BENCH 1.0
Leaderboard
Tasks
All benchmarks
Try in Browser
Blogs
Contribution
Gym
People
BENCHMARK
Benchmark
Benchmarks for Agents on a Virtual Filesystem.
01
VFS-Bench 1.0
Real tasks, each run on two interfaces: MCP and a virtual filesystem with VFS.md.
200 tasks · 5 models · 2000 runs