SEO testing: LLMs are bad at anchor selection but surprisingly good at picking target URLs
gaganghotra_ · x · 2026-10-23
SEO practitioner Gagan Ghotra tested most recently released LLMs on internal linking tasks: all models fall short at anchor text selection, yet they are surprisingly reliable at choosing the correct target URL even for weak anchors. He finds the asymmetry puzzling and is looking for explanations.
More from Models
- Kimi K3.1 Pro flagship model spotted rolling out on Kimi web and mobile — testingcatalog · 2026-10-24
- Qwen3.8-27B at 1.8bpw collapses to 13% on MiMo RL bench, EXL3 tests show — casper_hansen_ · 2026-10-24
- Ablating Jev-style decision models: LoRA-only hits 84.9, beating the full recipe — vanstriendaniel · 2026-10-24
- ByteDance's Dreamina offers full Seedance access for $0.29/month — can AI video rivals survive? — Proper_19 · 2026-10-23
- Claude's top failure mode: cites sources, then admits it never read them — dbreunig · 2026-10-23
- Dev praises Grok's client design: rare for an AI app to feel this considered — prasenx · 2026-10-23