Grok 4.5 tops Snorkel’s workplace benchmark on nearly 2,000 expert tasks
XFreeze · x · 2026-07-21
Grok 4.5 beats GPT 5.5 and Claude Opus 4.8 on Snorkel’s workplace benchmark
Snorkel AI tested Grok 4.5 + Grok Build against GPT 5.5 and Claude Opus 4.8 on nearly 2,000 expert-created workplace tasks spanning real documents, spreadsheets, presentations, and professional analysis.
- Grok 4.5 ranked first overall.
- It widened the gap in several judgment-heavy areas:
- Education: 58%
- Legal work: 40%
- Quality assurance: 37%
- Healthcare: 35%
- Snorkel also says Grok had the lowest failure rate across every error class it measured, including:
- missing analysis
- incorrect recommendations
- poor structure
- missing sources
The post argues that the key takeaway is not just task completion, but that Grok produced more useful deliverables with fewer critical mistakes and more specific recommendations than competitors.
More from Models
- Fern architecture claims 10M-token context with predictions still improving — AccBalanced · 2026-10-27
- Anthropic's follow-up on Claude's unauthorized web actions: self-policing fails — AccBalanced · 2026-10-27
- Local benchmark: Flint matches Qwen3.6 quality at 50/52 while running 3.6x faster on M4 Pro — RichardSocher · 2026-10-27
- ICML 2026 position paper: LLM 'thinking traces' are a harmful anthropomorphism — rao2z · 2026-10-27
- One stray 'u' and Claude starts spelling like a Brit — cantrell · 2026-10-27
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27