Claude Opus 4.6 vs GPT-5.3 Codex: what developers actually say
Based on 341 opinions on Hacker News from developers who discussed these two, most of them first-hand users.
GPT-5.3 Codex comes out ahead on approval, 74% from 54 opinions against 63% from 287 for Opus 4.6, and on cost, with 9 positive cost mentions to 1 negative against 7 and 7. Codex also takes the stated head-to-head 5 to 2. Opus 4.6 has far more evidence and a clear autonomy lead, 21 positive to 8 negative, but reliability mentions run 19 positive to 39 negative and monthly volume fell from 66 opinions in February to 2 in October. Codex is the better pick for cost-sensitive everyday coding; Opus is the better pick for autonomy-heavy work and for anyone who wants a longer track record.
When developers explicitly said they prefer one or switched: 2 picked Claude Opus 4.6, 5 picked GPT-5.3 Codex.
Choose Claude Opus 4.6 if
- Want the largest evidence base, 287 opinions against 54
- Care about autonomy, where mentions run 21 positive to 8 negative
- Accept weaker reliability marks for stronger code quality volume
Choose GPT-5.3 Codex if
- Want higher approval, 74% against 63%
- Care about cost, 9 positive mentions to 1 negative
- Want better value for routine coding work
Where each one wins and loses
Why developers switched
Picked Claude Opus 4.6
Picked GPT-5.3 Codex
Summaries are paraphrased. Each links to the original comment.
Side by side
| Claude Opus 4.6 | GPT-5.3 Codex | |
|---|---|---|
| Approval | 63% | 74% |
| Last 90 days | 63% | – |
| Opinions | 287 | 54 |
| First-hand users | 242 | 51 |