Announcement
Haiku 5.5 tops WebMCP benchmark’s combined success, speed and cost ranking
The benchmark’s author says Haiku 5.5 using WebMCP solved 49 of 49 tasks, versus 41 for GPT-6 Luna.
TLDR
The benchmark’s author says Anthropic’s Haiku 5.5 using WebMCP ranked first on its combined success, speed and cost score, ahead of GPT-6 Luna and previous winner Jev. Compared with the benchmark’s highest-ranked computer-use configuration, GPT-6 Luna, Haiku using WebMCP solved 49 of 49 tasks versus 41. The author reports 3.7 seconds per task versus 19.4, and a cost of 1/40 of a cent per task versus 0.21 cents.
Combined views
—
2 Sources, first seen ago
