{"slug": "opus-4-8", "title": "Opus 4.8", "summary": "Amp released Opus 4.8, a new AI coding agent that powers its 'smart' mode, replacing Opus 4.7. The model solves 62% of tasks in internal evaluations, up from 52% for its predecessor, by making more focused changes and verifying its own work more thoroughly. Opus 4.8 also offers a faster mode at 2.5× speed and 3× cheaper token costs.", "body_md": "Opus 4.8 now powers Amp's `smart`\n\nmode, replacing Opus 4.7.\n\nIt is a better coding agent than Opus 4.7: more faithful to the prompt, tighter in the changes it makes, and better at checking its own work. In our internal evals it solved **62%** of tasks, up from 4.7's **52%**.\n\nOpus 4.7 was already strong on hard, multi-file work, and Opus 4.8 keeps that. What changes is how it gets there: with fewer wasted steps and more self-checking.\n\nThe clearest difference is restraint and verification.\n\nOpus 4.7 can sometimes over-engineer, reaching for a more elaborate solution than the task needs. It also verifies its own work less, occasionally moving on even when a command's output is already warning that something is off.\n\nOpus 4.8 makes a more focused change that solves the specific intended task, then checks itself. It leans on a tighter write→test loop, often spinning up a quick script, test, or skill to confirm the change works before proceeding. In our evals it ran tests and code 15% more per task than 4.7.\n\nThat restraint is easiest to see on hard tasks. On everyday work the two make a similar number of tool calls. The harder a task gets, the tighter 4.8 stays, just where 4.7 tends to run long and fail more often.\n\nOpus 4.8 is noticeably better at using its tools and sub-agents without being told to.\n\nWhen a task needs outside context, it actually calls `librarian`\n\ninstead of inferring a library's behavior from the local code. Across our eval it reached for it 14 times, versus once for Opus 4.7. It also reaches for a repo's skills more often to verify its work, for example by driving the browser or the CLI, rather than just assuming the change worked.\n\nWhen it edits, it leans on `edit_file`\n\nfor surgical, in-place changes rather than rewriting whole files with `create_file`\n\n. In our evals 79% of its file edits go through `edit_file`\n\n, up from 63% on Opus 4.7.\n\nWe dropped the `Read`\n\ntool from `smart`\n\n.\n\nOpus 4.8 is good enough at reading files straight from the shell with `cat`\n\n, `rg`\n\n, `sed`\n\n, and `nl`\n\n. It parallelizes those reads when it needs several files at once.\n\nOpus 4.8 has a fast mode at roughly 2.5× the speed. It now costs 2× base tokens, down from 6× on 4.7—3× cheaper.\n\nToggle fast mode for a thread from the CLI command palette (`Ctrl+O`) → `speed: use fast`\n\n.\n\nOpus 4.8 stays close to what you ask, changes less to get there, and checks its own work. A few habits make it shine:", "url": "https://wpnews.pro/news/opus-4-8", "canonical_source": "https://ampcode.com/news/opus-4.8", "published_at": "2026-06-04 00:00:00+00:00", "updated_at": "2026-06-30 16:32:14.281879+00:00", "lang": "en", "topics": ["ai-agents", "large-language-models", "ai-tools", "ai-products", "developer-tools"], "entities": ["Amp", "Opus 4.8", "Opus 4.7"], "alternates": {"html": "https://wpnews.pro/news/opus-4-8", "markdown": "https://wpnews.pro/news/opus-4-8.md", "text": "https://wpnews.pro/news/opus-4-8.txt", "jsonld": "https://wpnews.pro/news/opus-4-8.jsonld"}}