Opus 5.5 Wins Developers Back from Codex, Early Reviews Say
Since its release in late September, Claude Opus 5.5 has drawn a wave of positive hands-on reviews centered on a "user comeback" narrative: users who previously switched to OpenAI Codex because of poor experiences with the last-gen Opus 5 are being pulled back to Anthropic by the new model. The current takeaway is that Opus 5.5 restores capability while offering better value, and several independent developers have made it their daily driver. This reputation reversal is worth watching closely, as it directly shapes the Claude-vs-Codex battle for the developer market.
Confirmed
- Every.to published a Vibe Check review of Opus 5.5; reviewer Kieran Klaassen said it has replaced Fable 5.1 as his daily driver model and is winning back users who had switched to Codex.
- On pricing, Every noted Opus 5.5 costs roughly 60% of Fable 5.1's token price (about a 40% discount) with comparable capability.
- Every team members who had previously moved to Codex over Opus 5's "personality issues"—argumentative, overly verbose explanations, drifting on small tasks—returned after the review, including Mike Taylor and Tyler Nishida, with Nishida expressing a positive verdict.
- Developer holdenmatt briefly called Opus 5.5 "the first good Opus since 4.8."
- Developer yawnxyz said non-Fable models have overtaken Fable in his usage for the first time, calling "opus 5.5 is cracked," with usage screenshots attached.
- @dejavucoder posted a long account of switching back to Anthropic from the GPT family: after about four months on the Codex app, he found GPT 5.6 Sol nearly all-capable, while the later Astra fell short on proactiveness and instruction following (the post is written interview-style with pseudonymous/fictional framing, so its reliability should be discounted).
Not yet confirmed
- All of the above are third-party reviews and personal experiences; Opus 5.5's specific capability benchmarks, version details, and official pricing strategy have yet to be confirmed by Anthropic.
- A large-scale Codex→Claude user回流 is so far only the judgment of reviewers and individual developers; whether it becomes a trend requires more data.
Why it matters
- Opus 5's interaction "personality issues" pushed some paying developers to Codex; Opus 5.5's rebounding reputation gives Anthropic a chance to win back lost ground.
- At roughly 60% of Fable 5.1's price with comparable capability, the value-for-money narrative—if validated at larger scale—could materially reshape model market pricing and selection.
2026-09-25 ~ 2026-09-27 · 6 related posts
Primary sources
- [source] Opus 5.5 Review: Every.to Says It's Pulling Codex Converts Back to Claude — every · 2026-09-25
- [source] Dev verdict on Opus 5.5: the first good Opus since 4.8 — holdenmatt · 2026-09-26
- Developer's detailed take: Opus 5.5 nears Fable 5.1 at lower cost, wins back Claude Code users — dejavucoder · 2026-09-26
- [source] Dev says Opus 5.5 is 'cracked', first time using it more than Fable — yawnxyz · 2026-09-27