On November 13 at 18:12 Singapore time, AI commentator Chubby (@kimmonismus, about 159K followers, editor-in-chief of Superintel) posted that after a year of switching between Codex/GPT and Claude Code/Opus, Claude Code with Opus 5.5 has been his daily driver for two months: "the most efficient and intelligent model to date," with GPT "good, but not close." He added that Claude's weak spot is computer use, where GPT is "unbeaten," and that he cares more about Claude Duet than ChatGPT dots, though neither is available to him in Europe. In a reply he put his split at about 80% Claude. At writing time the post had about 29K views, 590 likes and 152 replies. It went up during an Anthropic incident that blocked sign-ins to claude.ai and Claude Code from 09:58 to 11:58 UTC. This is one user's opinion and community discussion, not a benchmark.
Key Takeaways
- ✓The post drew about 29K views, 590 likes, 152 replies and 32 bookmarks; the author has about 159K followers and said in a reply he uses Claude about 80% of the time and GPT about 20%
- ✓Core claim: Opus 5.5 is "the most efficient and intelligent model to date" and GPT trails, but computer use is GPT's edge; OpenAI says GPT-6.1 Terra at max effort costs about 1/9 as much per task as Opus 5.5 (vendor claim)
- ✓Claude Duet's help center says the research preview is open only to US Pro, Max and Ultra users via a waitlist, and always runs on Opus 5.5
- ✓At writing time OpenAI's help center still says Pro dots are not offered in the EEA, Switzerland or the UK; the post's hint that Europe is getting access has no official announcement yet
- ✓Claude status page: sign-in errors on claude.ai and Claude Code from 09:58 UTC on November 13, a fix deployed at 11:05, resolved at 11:58, about two hours in total
Key Decision Metrics at a Glance
Heavy Claude Code use: compare subscription limits and API bills
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and pain points
This year's AI coding story has been Anthropic and OpenAI trading the lead: the GPT-6 series and Codex pulled many developers over, then Opus 5.5 brought a wave back to Claude Code. Many heavy users pay for both, so the question becomes which one is the daily driver and which jobs go where. On November 13, AI commentator Chubby (@kimmonismus, about 159K followers) gave his answer: Claude Code with Opus 5.5 has been his main tool for two months, and GPT is "good, but not close."
Architecture highlights
His argument has three parts. First, the model: Opus 5.5 is "the most efficient and intelligent." Second, the gap: computer use, meaning a model that reads the screen and clicks through desktop apps and browsers, where he says GPT is "unbeaten" and hopes Anthropic catches up. Third, products: he is more interested in Claude Duet, Anthropic's always-on agent app, than ChatGPT dots, but neither is open to him in Europe. Many replies back a split of Claude for code and GPT for computer use; others argue the two labs are roughly equal on intelligence and the gap is Claude Code's harness and UX; one reply says GPT-6 Astra is still steadier for complex builds such as a game engine from scratch.
Benchmarks and data
This is personal experience, not a benchmark. Official reference points: at launch OpenAI said GPT-6.1 Terra at max effort on Agents' Last Exam is over 3x as token efficient as Opus 5.5 at about 1/4 the price per token, for about 9x lower cost per task; that is a vendor claim and Anthropic has not responded. Duet's help center says it always runs on Opus 5.5 and is US only; OpenAI's help center says Pro dots are not offered in the EEA, Switzerland or the UK. The post had about 29K views, 590 likes and 152 replies.
Getting started
The post landed during a Claude sign-in outage (09:58 to 11:58 UTC) that locked users out of claude.ai and Claude Code, and replies used it as a reminder that relying on one vendor is risky; keep Codex or a local model ready as a fallback. In practice, route by task: in-repo refactors and long reasoning chains to Claude Code, workflows that click through web pages and desktop apps to GPT-6.1 Terra or Codex computer use. Comparing both on your own repo and failing tests is more reliable than anyone's subjective ranking.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.