芯片缺陷评测基准First Silicon发布，前沿AI检出率不足两成
Start here
2026年11月5日，研究团队正式推出面向硬件缺陷检测的评测基准“First Silicon”。相关测试结果显示，当前前沿AI模型在芯片硬件Bug检出能力上面临较大挑战。在评测中，表现最好的模型 Claude Opus 5.5 在零误报的严格标准下，仅能检出 18.6% 的真实芯片缺陷；同时，该模型发出的缺陷警报中约有三分之二为误报，表明前沿AI在严苛的芯片验证场景下仍存在明显的检出短板与高误报问题。
AI-generated from reports · Updated 3 hours ago
Report timeline
Follow the reports to see different sides of the story.
- Hacker News · AISelectedFirst Silicon Hardware Benchmark Released: Frontier Models Detect Under 20% of Chip Flaws at Zero False Positives
A research team has launched First Silicon, a hardware benchmark. Results show that top-performing Claude Opus 5.5 detected only 18.6% of real chip flaws under a strict zero-false-positive threshold, with two-thirds of its alerts being false alarms.
Story heat trend
Current heat 9·Comparable peak 10（Nov 5 02:00）·Comparable change, last 24 hours–
Trends compare only the same parties observed continuously, so the scope can be narrower than current heat counts. Move the pointer or tap the chart to see hourly heat; use the left and right arrow keys on a keyboard.