VFS-Bench 1.0 Leaderboard
Each model runs every task on both interfaces. Compare its success rate, tokens and cost with VFS and with MCP.
VFS vs MCP
Success Rate vs Cost
Up and to the left is betterSuccess Rate
PASS@1Higher is betterClaude Opus 5.5+4%
GPT-6.1 Sol0%
GPT-5.6 Luna+20%
Claude Haiku 5.5+6%
Gemini 3.8 Flash+4%
Average Tokens per Task
Lower is betterToken Reduction RateClaude Opus 5.573%
GPT-6.1 Sol72%
GPT-5.6 Luna60%
Claude Haiku 5.510%
Gemini 3.8 Flash25%