Claude Opus 5.5: what developers actually say

We read every Hacker News comment from Nov 2025 – Nov 2026 that gave a real opinion on Claude Opus 5.5. 148 of the 186 opinions come from people who say they use it themselves.

75%approval, all opinions

75%approval, last 90 days

#7of 45 models ranked

175different developers

Developers give Claude Opus 5.5 a 75% approval score across 186 opinions from 175 authors, and 148 of those opinions are firsthand. Code quality drives most of that: 56 positive mentions against 16 negative. Cost and autonomy also get more praise than complaints, while reliability is the one net-negative aspect at 4 positive against 9 negative. Sentiment slips across recent months, from 78 in September 2026 to 73 in October to 70 in November, though November holds only 10 opinions.

Pick it if

Want strong code quality on real projects
Want an agent that works with little supervision
Already work inside Claude Code

Look elsewhere if

Need predictable weekly usage
Expect stable behavior across long sessions
Documentation updates are a core task

Strengths and weaknesses

Each row counts opinions that mentioned the topic. Bars to the right of the line are praise, bars to the left are complaints.

PraiseComplaints

Code quality

Cost & pricing

Autonomy

Usage limits

Reliability

Speed

Large codebases

How opinion has moved

Monthly approval (line) and number of opinions (bars). Months with fewer than 3 opinions are left out.

What developers say

Praise

Says Opus 5.5 at high effort handles their multi-repo work and can drive phones via adb for testing.HN comment · Nov 2, 2026

Building a reliable IPTV app with Opus 5.5 works far better than anything else they have tried.HN comment · Nov 2, 2026

Says Opus 5.5 is unbeatable right now, with strong taste in the output it produces.HN comment · Nov 1, 2026

Says Opus 5.5 at medium effort is enough for most tasks, so high reasoning effort isn't needed.HN comment · Nov 1, 2026

Says Claude is the only option they trust for real work on large projects, unlike local models.HN comment · Nov 1, 2026

Says Opus 5.5 wrote an accurate PR description from minimal context, impressing them after earlier disappointing models.HN comment · Nov 1, 2026

Uses Opus 5.5 for most work and says it is in a league of its own.HN comment · Nov 1, 2026

Prefers Opus 5.5 for coding, partly from familiarity with Claude Code.HN comment · Oct 31, 2026

Complaints

Says Opus 5.5 misses details and needs rewrites, so humans are more trustworthy on some tasks.HN comment · Nov 1, 2026

Says Opus 5.5 refuses to write exploits against their own test infrastructure, citing safety.HN comment · Nov 1, 2026

Says Opus 5.5 insists on working alone rather than delegating to other agents for evaluation.HN comment · Oct 31, 2026

Says Opus 5.5 degrades over long sessions, forgetting instructions and previously given steps.HN comment · Oct 30, 2026

Says Opus 5.5 burns through tokens fast, exhausting a Max 20x weekly quota in a day or two.HN comment · Oct 29, 2026

Finds Opus 5.5 condescending and unwilling to explore unorthodox ideas, so prefers another model for brainstorming.HN comment · Oct 29, 2026

Hit a five-hour usage limit within 30 minutes on Opus 5.5 and found it much slower.HN comment · Oct 23, 2026

Says Opus 5.5 refused an SVG steganography prompt they wanted to test.HN comment · Oct 23, 2026

Summaries are paraphrased. Each links to the original comment.

Claude Opus 5.5 compared

Claude Opus 5.5 vs Claude Opus 5 Claude Opus 5.5 vs GPT-6 Astra Claude Opus 5.5 vs Claude Fable 5.1

FAQ

Is Claude Opus 5.5 worth it?

Approval sits at 75% from 186 opinions, 148 of them firsthand. Code quality is the strongest signal at 56 positive mentions against 16 negative. Cost follows at 27 positive against 6 negative.

Is Claude Opus 5.5 better than OpenAI Codex?

Head-to-head mentions are thin: one developer preferred Opus 5.5 over Codex and one preferred Codex. The clearer gaps are against Claude Opus 5 at 5 to 1 and GPT-6 Astra at 4 to 1.

Is Claude Opus 5.5 reliable?

Reliability is the only aspect with more complaints than praise, 4 positive against 9 negative. Users report degradation over long sessions and missed details that force rewrites.