Grok 4.7 Sweeps Top 3 on VulcanBench Frontier v4, Beating Opus 5.5 and Sonnet 5.5
XFreeze · x · 2026-10-28
A VulcanBench Frontier v4 leaderboard shared by X user XFreeze shows Grok 4.7 taking all top three spots: xHigh at 93.15, High at 92.71, and Medium at 92.30 — ahead of Haiku 5.5 (92.16), Sonnet 5.5 (92.02), Fable 5.1 (91.84), and Opus 5.5 (91.11).
The benchmark targets repository-level software engineering rather than isolated coding tasks: 23 behavioral-reconstruction tasks where models must rebuild legacy software to match real behavior, scored on functional correctness, hidden tests, security, lint/complexity, and code quality. The top Grok 4.7 run also posted the highest code quality score of 84.35, passing 23/23 tasks.
Note: the leaderboard comes from a third-party account; the model version is not yet officially confirmed.
More from coding & agent
- Claude built its own Obsidian plugin, and its human is reviewing AI-edited videos on iPhone — evielync · 2026-10-28
- One illustration to a full VTuber model: dev builds open-source kit using Claude and Codex agents — bdsqlsz · 2026-10-28
- Ahrefs Podcast: How Ryan Law Vibe-Coded a Full Website in One Day — 5le · 2026-10-28
- Dev explains Microsoft's four Copilots as nested if, for, while and while(true) — pswider · 2026-10-28
- Developers saving 10+ hrs/week with AI fell from 68% to 25% in a year — Suspicious_Orchid770 · 2026-10-28
- 10 MCP servers in Docker: tool counts range from 1 to 46 per server — LeastAd2616 · 2026-10-28