Opus 5.5 Launch Wave: Users Hail a Step Change as Codex Users Return

Claude Opus 5.5 launched on September 24 and drew a rare wave of unanimous praise on X and Reddit over the following two days, with multiple users saying its reputation has reversed sharply versus its predecessor—even Anthropic engineers jumped in to respond. All conclusions so far come from early hands-on impressions; performance beyond officially demonstrated cases still needs longer validation.

Confirmed

• User kimmonismus called Opus 5.5 the best model he has ever used: great taste, fast, smart, and finally precise without rambling. Having fully switched to Codex before, he has now returned to Anthropic and says Opus 5.5 directly replaces Astra. Anthropic engineer Boris Cherny replied in agreement, and Lydia Hallie also proactively collected day-one feedback.
• User RileyRalmuto called it the "craziest" jump since Opus 4.5, a paradigm-shifting step change. Netizen tunahorse21 (relayed by BLUECOW009) said it was the biggest capability leap they had felt, and they never opened Codex again that day.
• Multiple users confirmed improvements in writing style and interaction: marktenenholtz said replies are clear and direct, no longer outputting agent-style boilerplate; vista8 said the model understands quickly, has its own judgment, and often delivers solutions better than the original request (e.g., proactively adding icons to a menu); @bitbanter (via repost) said they can finally delete their custom instruction "don't write like an idiot"; tacomaster05 said it read 30,000-to-50,000-word documents in one pass and remembered nearly everything, pleading with Anthropic not to nerf it.
• gleech shared initial data: zero refusals across a hundred sessions, and writing tests showed it clearly better than the previous Fable. His crowdsourced site Model Behavior Reports already has 67 reports related to Opus 5.5 (2,116 reports across 14 models site-wide).
• The Every team published a Vibe Check test: Opus 5.5 pulled back several former heavy Claude users who had switched to Codex, with capability close to Fable 5.1 at roughly 40% of its token price (a 60%-discount level), no longer over-engineering small tasks and not arguing at the drop of a hat.
• User Ice2jc ran a side-by-side test: asking Astra and Opus 5.5 to debug project issues simultaneously, Opus 5.5 admitted Astra's solution was better in 5 out of 5 attempts over one afternoon.
• Other positive feedback: weswinder said usability has rebounded substantially from Opus 5 and icon generation detail is stunning; ayushtweetshere said its performance convinced him to cancel his cancellation and stay (while criticizing Anthropic's double-standard safety marketing of painting models as dangerous while releasing them); TheOyinbooke bluntly said "OpenAI has something to chase."

Unconfirmed

• All of the above comes from the first two days post-launch. gleech cautioned that a model's real weaknesses often only show up after two weeks of sustained use; weswinder, tunahorse21, and others are sharing personal impressions, and performance still awaits more reviews.

Why It Matters

• The poor reception of Opus 5 and its predecessor drove users to OpenAI Codex. This round of reviews centers on "capability leap + writing style back on form + pricing advantage," directly reversing the user outflow. The pricing and product pressure on competitors (as TheOyinbooke put it) is the core story to watch here.

2026-09-24 ~ 2026-09-25 · 22 related posts

Full story(16 episodes)→

• Episode 1: Claude Opus 5.5 Spotted in Claude Code Ahead of Imminent Launch(2026-09-23, 8 posts)
• Episode 2: Anthropic Launches Claude Opus 5.5: Stronger, 40% Cheaper, Tops AA Index(2026-09-23, 259 posts)
• Episode 3: Anthropic Launches Banked Usage Resets Alongside Opus 5.5(2026-09-23, 10 posts)
• Episode 4: Perplexity Rolls Out Opus 5.5 to All Users, Citing 67.6% Cost Savings(2026-09-23, 2 posts)
• Episode 5: Anthropic Teases Sonnet 5.5 and Haiku 5.5 Within Weeks(2026-09-23, 2 posts)
• Episode 6: Claude Code 2.1.280 Ships with Opus 5.5 as Default Model(2026-09-23, 3 posts)
• Episode 7: Claude Opus 5.5 Rumored to Fall Back to Weaker Models on Frontier Dev Tasks(2026-09-23, 4 posts)
• Episode 8: Claude Opus 5.5 Sparks Wave of Community Demos Within 24 Hours(2026-09-23, 14 posts)
• Episode 9: Dev recreates Minecraft in browser with Opus 5.5(2026-09-23, 2 posts)
• Episode 10: Rumor: Claude Opus 5.5 Stuns OpenAI, GPT-6.1 Timeline Squeezed(2026-09-23, 3 posts)
• Episode 11: Opus 5's FrontierCode Score Drops from 53.4% to 48% in New Card(2026-09-23, 3 posts)
• Episode 12: Developers Say Opus 5.5 Is Faster, Cheaper and Back as Daily Coding Model(2026-09-23, 4 posts)
• Episode 13: Claude Opus 5.5 Shown Generating Pixel-Art Animated Scenes(2026-09-23, 2 posts)
• Episode 14: Opus 5.5 Launch Wave: Users Hail a Step Change as Codex Users Return(2026-09-24, 22 posts)
• Episode 15: Claude Opus 5.5 Tops Code Arena, Beating GPT-6 Astra by 26 Points(2026-09-24, 2 posts)
• Episode 16: Claude Opus 5.5 Delivers Fast, Proactive Coding That Developers Say Enables Flow State(2026-09-24, 2 posts)

• User cancels switch to Codex after Opus 5.5's 40% price cut wins him back — ayushtweetshere · 2026-09-24
• Claude Engineer Reacts as Early Opus 5.5 Access Reportedly 'Blew Away' Expectations — felixrieseberg · 2026-09-24
• [source] First-day Opus 5.5 verdict: power user says it replaced Astra entirely — kimmonismus · 2026-09-24
• Early user feedback: Opus 5.5 is 'finally a usable Opus again' — weswinder · 2026-09-24
• Early User: Opus 5.5 Feels Like Opus 4.6 but Answers Clearly Without Agent-Slop — marktenenholtz · 2026-09-24
• Blogger Finds Opus 5.5 Impressively Smart, Often Suggests Better Solutions Than Asked — vista8 · 2026-09-24
• Opus 5.5 Shows Off UI Flair: Checkmark Icons and Menu Polish on Its Own — vista8 · 2026-09-24
• [source] Opus 5.5 vibe check: 60% cheaper than Fable and winning back Codex converts — every · 2026-09-24
• Opus 5.5 early test: zero refusals in 100 sessions, better writing — gleech · 2026-09-25
• Model Behavior Reports crowds 2,116 LLM behavior reports across 14 models — gleech · 2026-09-25
• Early verdict: Opus 5.5 is the best Claude yet — well-rounded and faster, weaker at math — marcosalvi · 2026-09-25
• Opus 5.5 Wins Back Codex Converts: Every.to Team Vibe Check — every · 2026-09-25
• Dev runs Opus 5.5 vs Astra cross-audit workflow, Opus concedes 5 times in one afternoon — Ice2jc · 2026-09-25
• Early hands-on: Opus 5.5 generates icons with 'insane' attention to design detail — weswinder · 2026-09-25
• User says Opus 5.5 writes so well he deleted his 'don't write like a fuckhead' custom instruction — generativist · 2026-09-25
• Users say Opus 5.5 is the biggest coding model jump in a long time — BLUECOW009 · 2026-09-25
• User: Opus 5.5 Is Surprisingly Good at Making Videos — idanbeck · 2026-09-25
• Early hands-on: RileyRalmuto calls Opus 5.5 a step change since Opus 4.5 — RileyRalmuto · 2026-09-25
• [source] Blogger Hails Opus 5.5 as Best Model Yet: Tasteful, Precise, Not Verbose — kimmonismus · 2026-09-25
• User says Opus 5.5 reads 30k-50k word docs almost perfectly, begs Anthropic not to nerf it — tacomaster05 · 2026-09-25
• Anthropic Engineer Weighs In as Users Say Opus 5.5 Won Them Back from Codex — bcherny · 2026-09-25
• "Sorry OpenAI, you have a lot of catching up to do" — Opus 5.5 praise — TheOyinbooke · 2026-09-25