DevConsensus
Comparisons / Claude Opus 5 vs Claude Opus 4.8
AI models for coding · HN opinions Nov 2025 – Nov 2026

Claude Opus 5 vs Claude Opus 4.8: what developers actually say

Based on 296 opinions on Hacker News from developers who discussed these two, most of them first-hand users.

Claude Opus 5
33%
approval · 162 opinions · 148 developers
last 90 days: 30%
Claude Opus 4.8
49%
approval · 134 opinions · 128 developers
last 90 days: 57%

B (Opus 4.8) comes out ahead. It has higher approval (49 vs 33), higher 90-day approval (57 vs 30), a 7 to 2 edge in stated head-to-head preferences, and positive code quality sentiment (35 positive, 27 negative) where A is negative (24 positive, 36 negative). A's one clear edge is cost, where sentiment is slightly positive (4 positive, 3 negative) against B's negative split (6 positive, 10 negative), and A also has a small speed edge (3 positive, 2 negative vs 1 and 4). Pick B for consistency and daily use, and pick A only when cost or a newer model matters more than reliability (1 positive, 26 negative), UX (1 positive, 20 negative), and A's sharply negative October trend (6 positive, 75 negative).

When developers explicitly said they prefer one or switched: 2 picked Claude Opus 5, 7 picked Claude Opus 4.8.

Choose Claude Opus 5 if

  • Want a small cost and speed edge
  • Can accept weaker reliability and UX scores
  • Want the newer model despite lower approval

Choose Claude Opus 4.8 if

  • Want higher approval and steadier code quality
  • Care about UX and reliability
  • Weight the 7 to 2 head-to-head split

Where each one wins and loses

PraiseComplaintsA = Claude Opus 5B = Claude Opus 4.8
Code quality
A
36
24
B
27
35
Reliability
A
26
1
B
20
7
UX
A
20
1
B
3
1
Autonomy
A
12
5
B
3
2
Cost & pricing
A
3
4
B
10
6
Speed
A
2
3
B
4
1
Usage limits
A
3
B
3
2
Large codebases
B
2
1

Why developers switched

Picked Claude Opus 5

Picked Claude Opus 4.8

Says Opus 5 was so bad in real use that they switched back to Opus 4.8.HN comment · Sep 23, 2026
Prefers Opus 4.8 at work because Opus 5 is too verbose and Opus 4.8 is smart enough.HN comment · Sep 12, 2026
Says GLM 5.3 resembles Opus 4.8, which they consider better than Opus 5 and less chatty than DeepSeek V4.HN comment · Aug 29, 2026
Suspects Anthropic weakened Opus 5 to widen the gap with Fable, so they stayed on Opus 4.8.HN comment · Aug 23, 2026
Says Opus 5 swings between much better and much worse than Opus 4.8, so they prefer 4.8's consistency.HN comment · Aug 23, 2026
Calls Opus 5 a downgrade from 4.8, citing poor instruction following, scope creep, and bloated prose and comments.HN comment · Aug 19, 2026

Summaries are paraphrased. Each links to the original comment.

Side by side

Claude Opus 5Claude Opus 4.8
Approval
33%
49%
Last 90 days30%57%
Opinions162134
First-hand users136117

Related comparisons