<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/DSxXJVBPi" -->

---
title: Opus 5.5 is 'the greatest model of all time,' says Theo
description: Developers on social media are split over the Opus 5.5 model. Theo calls it 'the greatest model of all time' and says of anyone less impressed that 'you still...
canonical: https://daily.dev/posts/opus-5-5-is-the-greatest-model-of-all-time-says-theo-dsxxjvbpi
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Opus 5.5 is 'the greatest model of all time,' says Theo | daily.dev
og:description: Developers on social media are split over the Opus 5.5 model. Theo calls it 'the greatest model of all time' and says of anyone less impressed that 'you still...
og:url: https://daily.dev/posts/opus-5-5-is-the-greatest-model-of-all-time-says-theo-dsxxjvbpi
og:image: https://api.daily.dev/og/posts/DSxXJVBPi.png
og:image:alt: Opus 5.5 is 'the greatest model of all time,' says Theo
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Opus 5.5 is 'the greatest model of all time,' says Theo

**[Trends](https://daily.dev/sources/trends)** · 3 min read · 5 upvotes · 0 comments

## Summary

Developers on social media are split over the Opus 5.5 model. Theo calls it 'the greatest model of all time' and says of anyone less impressed that 'you still haven't shipped anything impressive with it.' Scaling01 posted just 'Opus 5.5 supremacy.' Swyx calls it 'the best work estimator model' he's ever used, and says calibration matters more as we delegate work. Alexpowelldc is more cautious, keeping workflows model agnostic, and says he gave up on the earlier Opus 5 because it was 'awful to work with.'

## Content

Opus 5.5 has been out for over a month, and the discourse has split into worship, shrugs and flat contempt.

## The worshippers

@theo is the loudest: "Opus 5.5 is the greatest model of all time and I'm still struggling to find the ceiling." He also says dropping from extra high to high effort is "better almost all of the time." @scaling01 posted "Opus 5.5 supremacy." @swyx calls it the best work estimator he's used, because it doesn't quote a human's 2-3 days for a job it can finish in 2-3 hours. @trikcode says it gives him more days off without taking his job, and begs Anthropic to "please just stop here." The same user thought the model had been nerfed, then checked a daily tracker and found it scoring slightly higher than at launch.

The numbers lean the same way. Opus 5.5 leads a Chinese "From Top to Basic" leaderboard with 58 points, and @Hesamation points to an Artificial Analysis personality index where it beats GPT 6.1 Sol 93 to 18 on being informative. @OfirPress congratulated Anthropic on a new ProgramBench high score.

## The skeptics

@yacineMTB isn't having it: "Opus 5.5 is no where near as smart as I am. It is RL maxed dogshit." It can't write good code unless you have neovim open beside it, he says, and "if you have any other experience, you are a bad programmer." @theo also posted something that reads like a jab at the rest of us: "It's been over a month since Opus 5.5 dropped and you still haven't shipped anything impressive with it." @ryanflorence is wondering if he's falling behind. His feed is full of "opus made this mini game" posts while his own run chewed through 8 hours producing markdown for combat and skill-tree design.

## The showpieces

Those games are real. Simon Willison says October is another inflection point and that Opus 5.5 can build video games that are actually fun. He shipped four browser games, each under four hours of his own time, and says GPT-6 Astra's sense of fun doesn't match yet.

Then there's Karpathy's HackerTube. Per his blog, Claude pulled Hacker News comments, had Sonnet filter them, and built a site of 6,454 recommended videos in about two hours. He still bought the domain and booted the VPS himself. One commenter noted the manual steps are the real story, since API keys and anything near production still go through a human.

@OfirPress also reshared a post claiming Opus 5.5 made OpenAI-style promo films in pure code with three.js and GLSL.

## Where it lands

The split looks less like capability and more like workflow. The loudest fans are doing greenfield work, games and demos. The loudest critics are working in real codebases, where @theo himself says a competing model hallucinates "egregiously" and is "10x worse." That's a complaint about GPT, but it shows where the bar sits.

## Questions this post answers

### what are developers saying about Claude Opus 5.5 compared to Opus 5

Reactions are largely positive but split in tone. Theo calls it 'the greatest model of all time,' swyx calls it 'the best work estimator model' he's used because it gives realistic AI-speed timelines instead of human ones, and alexpowelldc says he abandoned Opus 5 because it was 'awful to work with,' suggesting some praise reflects relief after a weak prior version.

_Developers weighing whether to upgrade their coding agent can track model reactions like these on daily.dev._

## Community take

How the wider developer community reacted, aggregated from 4 discussions and 39 comments across x (as of 2026-11-02).

**TL;DR:** Most of the visible reaction focuses narrowly on the HackerTube demo, with people impressed by the 43M-comment filtering pipeline and the two-hour build time rather than debating Opus 5.5's broader coding ability.

**Sentiment:** 70% positive · 25% mixed · 5% skeptical

**The case for**

- The filtering pipeline (regex/URL matching before any model call) is seen as a smart, cheap way to cut 43M comments down to a manageable set.
- Swapping clickbait thumbnails for a middle-of-video frame is repeatedly praised as a clever UX touch.
- People like that an agent handled the 'boring middle' (API key, server, domain, deploy) that normally kills side projects.

**The pushback**

- One reply notes the curation only works because the person running it already has good taste; the same pipeline would just produce a faster feed of clickbait for someone without that judgment.
- A reply jokes that closing the tab took longer than building the two-hour demo, implying the output isn't necessarily something people keep using.

**By community**

- x (positive): Replies are largely enthusiastic about the HackerTube demo's speed and clever filtering, with only light skepticism about taste-dependence and lasting usefulness.

**Open questions**

- Whether the curation quality holds up past the top few thousand filtered videos or degrades in the long tail.
- Which model actually did the comment filtering versus the video-site build (Opus vs. Sonnet).

**Highlights**

> @kloss_xyz Worth knowing how he kept it cheap: Claude didn't read all 43M comments. Most got cut first with URL matching, regexes and spam drops, and only the survivors went to Sonnet via a Claude Code workflow on his subscription tokens. The model was the last step, not the pipeline.
> — [chasers24k on x · 1 points](https://x.com/chasers24k/status/2116987314199003635)

> @kloss_xyz The intresting part isn't the filtering, it's who sets the taste. Karpathy's filter works because he already knows what good looks like. Hand the same agent to someone without that and you just get a faster feed of the same clickbait.
> — [Kieren\_PB\_Day on x](https://x.com/Kieren_PB_Day/status/2116997156485955713)

> @rohanpaul_ai The reusable idea is the filter: count what engineers keep recommending to each other across 20 years of comments, instead of trusting a ranking algorithm. That works on any forum or Discord archive you can export.
> — [Cozy2054934 on x](https://x.com/Cozy2054934/status/2117017693975175460)

> @rohanpaul_ai The wild part isn't the code, it's the glue: API key, VPS, domain, deploy. Projects used to die in that boring middle. When an agent eats the boring middle, 'someday' projects become 'before breakfast' projects.
> — [JO0Ie on x](https://x.com/JO0Ie/status/2117060082294759555)

> @benny1388 Two hours to build HackerTube; considerably longer to close the tab.
> — [thesimaximalist on x](https://x.com/thesimaximalist/status/2117071804409987526)

**Source threads**

- [x](https://x.com/rohanpaul_ai/status/2116983513425883148) · 0 points · 15 comments
- [x](https://x.com/benny1388/status/2117057677390160016) · 3 points · 3 comments
- [x](https://x.com/MakingCentsnow/status/2116972152888922382) · 1 points · 0 comments
- [x](https://x.com/kloss_xyz/status/2116985393845600634) · 0 points · 21 comments

## Similar posts on daily.dev

- [Anthropic releases Opus 4.5 with new Chrome and Excel integrations](https://daily.dev/posts/anthropic-releases-opus-4-5-with-new-chrome-and-excel-integrations-o27v4qrok) · TechCrunch · 3 upvotes · 0 comments

---

Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#llm](https://daily.dev/tags/llm), [#claude](https://daily.dev/tags/claude), [#anthropic](https://daily.dev/tags/anthropic)

[View this post on daily.dev](https://daily.dev/posts/opus-5-5-is-the-greatest-model-of-all-time-says-theo-dsxxjvbpi)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Opus 5.5 is 'the greatest model of all time,' says Theo","url":"https://daily.dev/posts/opus-5-5-is-the-greatest-model-of-all-time-says-theo-dsxxjvbpi","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/opus-5-5-is-the-greatest-model-of-all-time-says-theo-dsxxjvbpi"},"datePublished":"2026-10-26T04:51:02.484Z","dateModified":"2026-11-02T04:33:00.779Z","description":"Developers on social media are split over the Opus 5.5 model. Theo calls it 'the greatest model of all time' and says of anyone less impressed that 'you still...","isAccessibleForFree":true,"articleSection":"Trends","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/opus-5-5-is-the-greatest-model-of-all-time-says-theo-dsxxjvbpi","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":5},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,llm,claude,anthropic","timeRequired":"PT3M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"Opus 5.5 is 'the greatest model of all time,' says Theo"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/opus-5-5-is-the-greatest-model-of-all-time-says-theo-dsxxjvbpi#faq","mainEntity":[{"@type":"Question","name":"what are developers saying about Claude Opus 5.5 compared to Opus 5","acceptedAnswer":{"@type":"Answer","text":"Reactions are largely positive but split in tone. Theo calls it 'the greatest model of all time,' swyx calls it 'the best work estimator model' he's used because it gives realistic AI-speed timelines instead of human ones, and alexpowelldc says he abandoned Opus 5 because it was 'awful to work with,' suggesting some praise reflects relief after a weak prior version. Developers weighing whether to upgrade their coding agent can track model reactions like these on daily.dev."}}]}
```

