Claude Opus 4.8: what developers actually say

We read every Hacker News comment from Nov 2025 – Nov 2026 that gave a real opinion on Claude Opus 4.8. 117 of the 134 opinions come from people who say they use it themselves.

49%approval, all opinions

57%approval, last 90 days

#39of 45 models ranked

128different developers

Developers are split on Claude Opus 4.8. Approval is 49 across 134 opinions, and 57 over the last 90 days. Code quality draws the most praise at 35 positive against 27 negative, while reliability is the weak spot at 7 positive against 20 negative. Monthly approval fell to 38 in June, recovered to 50 in July, and the recent rise rests on small samples of 2 to 9 opinions.

Pick it if

Want strong code review and bug catching
Need a model that follows written project rules
Prefer concise answers over long explanations

Look elsewhere if

Need steady reliability across long agent runs
Work in large or complex codebases
Have tight usage limits or a small budget

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

Reliability

Cost & pricing

Speed

Autonomy

Usage limits

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

Calls Opus 4.8 their most useful coding model.HN comment · Sep 13, 2026

Uses Opus 4.8 at work; says it is smart enough and less verbose than Opus 5.HN comment · Sep 12, 2026

Says Claude Opus 4.8 consistently catches issues Gemini misses in review.HN comment · Sep 3, 2026

Says GLM 5.3 feels like Opus 4.8, which they consider better than Opus 5.HN comment · Aug 29, 2026

Says they prefer Opus 4.8 because its performance is more consistent.HN comment · Aug 23, 2026

Says Opus 4.8 and Fable are the only models that aren't frustrating to use.HN comment · Aug 12, 2026

Says Opus 4.8 followed their CLAUDE.md rules nearly always and never looped like this.HN comment · Jul 28, 2026

Says Opus 4.8 is plenty for their work before they moved to Opus 5.HN comment · Jul 28, 2026

Complaints

Says Opus 4.8 now has security-related restrictions that force starting new sessions.HN comment · Oct 30, 2026

Spent hours with Opus 4.8 on Stockfish game analysis and it never worked usefully.HN comment · Sep 26, 2026

Says after switching to Opus 4.8, their adversarial agents argued heavily and filled the code with stale comments.HN comment · Sep 14, 2026

Says Opus 4.8 suddenly regressed after going well, ignoring their design guide and inlining styles.HN comment · Aug 22, 2026

Was frustrated by Opus 4.8's many mistakes when accidentally returned to it.HN comment · Aug 14, 2026

Says 4.8 basically could not handle their complex full-stack project.HN comment · Aug 12, 2026

Says one call on Opus 4.8 uses 20-30% of the 5-hour window, making it barely usable.HN comment · Aug 6, 2026

Says Opus 4.8 has recently gotten much worse, with more hallucinations and comprehension errors.HN comment · Jul 29, 2026

Summaries are paraphrased. Each links to the original comment.

Claude Opus 4.8 compared

Claude Opus 4.8 vs GPT-5.5 Claude Opus 4.8 vs Claude Opus 5 Claude Opus 4.8 vs Claude Fable 5 Claude Opus 4.8 vs Claude Opus 4.6

FAQ

Is Claude Opus 4.8 worth it?

It depends on the task. Code quality scores well, with 35 positive mentions against 27 negative. Reliability is the main concern at 7 positive against 20 negative, so test it on your own work first.

Is Claude Opus 4.8 better than GPT-5.5?

Opinions lean toward GPT-5.5. Twelve developers said they preferred GPT-5.5 instead, while 8 preferred Opus 4.8. Several users still keep Opus 4.8 for review and planning.

Is Claude Opus 4.8 reliable?

That is its weakest area. Reliability drew 7 positive and 20 negative mentions. Some users report regressions and more hallucinations, while others say its performance is consistent, so experiences differ.