Claude vs ChatGPT vs Gemini: Opus 5.5, GPT-6 Astra and Gemini 4 Argon on the Same Builds

Claude is the one to pick for hard build work. Anthropic's Opus 5.5 placed first in all four builds of Pat Simmons's Gemini 4 Argon review, which he ranked blind: a photoreal 3D cold brew bottle, a product launch video, a car soccer game and a rebuild of a 3D-heavy website. ChatGPT's top model, GPT-6 Astra, placed above Google's Gemini 4 Argon every time, and Argon came last in all four.

Gemini was the cheapest to run. Argon's 3D bottle cost $6.85 at API prices, against $8.99 for Astra and $29.95 for Opus 5.5, which took 114 minutes to Argon's 36. In the apps, each company's top model starts on a plan of about $20 a month: Claude Pro ($20 billed monthly) for Opus 5.5, ChatGPT Plus ($20) for Astra and Google AI Pro ($19.99) for Argon.

Last checked November 5, 2026: plan prices and which plan includes each top model, on Anthropic's, OpenAI's and Google's plan pages, plus Argon's sale price on Google's API pricing page.

Which is best: Claude, ChatGPT or Gemini?

For long, hard builds, Claude did best. Opus 5.5 won every round of the Argon review, and Claude's other model in the test, Fable 5.1, placed second or third each time. After the launch video round Pat said, "Opus 5.5 is on a generational run right now with these model tests."

The same order holds across Pat's earlier rounds. In six same-prompt rounds between September 23 and October 25, 2026, Opus 5.5 came first in 16 of 21 builds, Fable 5.1 won four and GPT-6 Astra won one, as tallied in the best AI for coding, ranked from 21 builds.

ChatGPT came second of the three apps. GPT-6 Astra placed above Gemini 4 Argon on all four builds, and the two-way results are in Gemini 4 Argon vs GPT-6 Astra on the same 4 builds.

Here is how that plays out by job:

• Claude (Opus 5.5) for 3D, video and website builds where quality comes first.
• ChatGPT (GPT-6 Astra) as the runner-up, ahead of Gemini in every round.
• Gemini (Argon) when the cost per build is what you watch.

These results cover build work only. Pat didn't test everyday chat, writing, research or how accurate each app's answers are, and in the review he called his builds "very limited tests." At the end of the video he said it still made sense for him to use the Claude models or Astra.

How the three did on the same builds

Opus 5.5 placed first on all four builds and Gemini 4 Argon placed last on all four. Claude's Fable 5.1 and ChatGPT's GPT-6 Astra split second and third.

Ranks are Pat's blind picks out of four models, from his Gemini 4 Argon review, published October 14, 2026. The bottle and launch video ranks are from the reveals in the video above. The game and website ranks are from our Gemini vs ChatGPT results and the full Gemini 4 Argon review, with every prompt and live build.

A photoreal 3D cold brew bottle you can spin, open and pour

Each model had to build a bottle for a made-up coffee brand, Clanker Coffee, as a photoreal 3D product in the browser, with a glass of ice and a pour you trigger with a click. The models built it headless in Blender (the free 3D program) by writing Python scripts.

Opus 5.5 came first, Fable 5.1 second, Astra third and Argon fourth. When the ranking was revealed, Pat said, "Opus just crushes 3D work." Argon went for an illustrated look when the prompt asked for photoreal, and Pat said "the 3D was not realistic," though he added it was "still not a bad job overall from Argon."

A product launch video for the same coffee brand

The second job was a launch video, with photoreal product shots, music and sound. Every model got Pat's launch video skill, GPT Image Gen 2.5 for still images and Blender for 3D.

Opus 5.5 came first again, then Astra, Fable 5.1 and Argon, and Pat said it was close between Fable 5.1 and Argon. He singled out how Opus's video opens on a "hit enter" idea and cuts to a robot's eyes lighting up, and a camera swipe across the coffee bags with motion blur: "I've yet to see that from a model."

A car soccer game and a website rebuild

The last two jobs were a playable Rocket League-style car soccer game in the browser and a pixel-for-pixel rebuild of the oryzo.ai website, 3D included. Opus 5.5 won both and Argon came last in both.

Fable 5.1 placed second on the game and Astra second on the website. Astra kept working on the website for 499 minutes and ran up $357.84, the most expensive single build in the review, per the Argon review's cost tables.

Which one costs the least to run?

Gemini 4 Argon cost the least on every build. Across all four builds Argon came to $39.54 at API prices, against $189.04 for Opus 5.5, $203.08 for Fable 5.1 and $409.96 for GPT-6 Astra, per the cost tables in the Argon review.

Costs, tokens and times are from the cost cards Pat showed after each reveal in his Gemini 4 Argon review, published October 14, 2026. Tokens are the small chunks of text a model reads and writes.

These are API-equivalent costs: each build's real token count priced at the model's published per-token rate. Pat ran the builds on subscriptions. On Google's $99-a-month plan he hit Argon's daily usage limit after about one build, the 3D bottle, and said its usage goes "a little faster than Codex or Claude Code."

Argon's low bill comes partly from a sale. Google's API pricing page lists Argon at $2 per million input tokens and $10 per million output through November 30, 2026, then $4 and $20 from December 1. Anthropic's pricing page lists Opus 5.5 at $4 and $20, the same as Argon's price after the sale.

Which plan gets you each top model

Each company's top model in Pat's builds starts on its roughly $20 plan: Claude Pro for Opus 5.5, ChatGPT Plus for GPT-6 Astra and Google AI Pro for Gemini 4 Argon. None of the three free plans includes it.

Prices are US monthly prices from Claude's pricing page, ChatGPT's pricing page and Google's Gemini plans page, all checked November 5, 2026. More on Claude's free plan is in is Claude free.

Is Gemini still the best AI?

In Pat's build tests it wasn't: Gemini 4 Argon, the top model in the Gemini app, came last in all four builds of his review, behind both Claude models and GPT-6 Astra. Google leaned on 3D work in its launch demos, and Pat said Argon's own 3D build "wasn't even freaking close to their blog posts."

Google's own benchmarks put Argon first on some office and legal work. Google's Argon post says it ranks first on AutomationBench, Zapier's test of business tasks, with 51.3%, and Google's benchmark chart shows 19.6% on Harvey's Legal Agent Benchmark (the chart is in our Argon review). The same chart has Argon behind Opus 5.5 on Terminal-Bench, a coding test that Pat said usually matches day-to-day coding work. These are Google's numbers, and Pat's view of them in the review was that "these labs just love finding benchmarks where they perform the best."

Argon is in the Gemini app for Google AI Pro and Ultra subscribers. For what has changed in each app since, see what's new in Gemini, what's new in Claude and what's new in ChatGPT.

Are Claude and Gemini the same?

They are separate products from separate companies. Claude is made by Anthropic and Gemini by Google, and ChatGPT comes from OpenAI. Each company trains its own models and sells them on its own plans.

In each app you pick from that company's models. Claude's pricing page lists Opus, Sonnet and Haiku, plus Fable on paid plans. ChatGPT's lists GPT-6 Luna, Sol and Astra, and Google's Gemini plans list Flash-Lite, Flash, a Pro model and Gemini 4 Argon.

They can work with the same files. Pat noted in the review that Claude and ChatGPT can now reach Google Docs, Sheets and Slides through connectors, so he wasn't sure what advantage Gemini has left there. Claude's version is in Claude now works inside Google Docs, Sheets and Slides.

ChatGPT vs Gemini vs Copilot: where Copilot fits

Copilot is Microsoft's assistant, which Microsoft's plans page calls "the AI for work," built into Word, Excel, PowerPoint and Outlook. Pat hasn't tested Copilot, so it has no place in the rankings above.

Copilot has limited free use with a Microsoft account, per Microsoft's plan comparison. More use comes bundled with Microsoft 365, per Microsoft's Copilot plans page (checked November 5, 2026): Personal at $9.99 a month, Family at $12.99, Premium at $19.99 with "extensive" Copilot usage and Pro at $99.99 with the highest usage.

Microsoft is also putting real Word, Excel and PowerPoint files inside the Copilot app, starting with companies in its Frontier early-access program, as covered in Microsoft puts Word, Excel and PowerPoint inside Copilot. The tool page is Microsoft Copilot.

For other alternatives to Gemini and ChatGPT, including Grok, Meta AI and Perplexity, see ChatGPT alternatives and what each one costs.

How Pat tested them

Every model got the exact same prompt for each build, with no steering, and ran on high effort. Pat ranked each round blind, with the builds labeled A to D, and only then revealed which model made which.

The four models were Claude's Opus 5.5 and Fable 5.1, ChatGPT's GPT-6 Astra and Gemini 4 Argon. They worked as coding agents, writing code and driving Blender with Python scripts, and the launch video round added GPT Image Gen 2.5 and Pat's launch video skill. Every prompt and every live build is in the Gemini 4 Argon review.

Pat's earlier rounds without Gemini are in his Opus 5.5 review and GPT-6 Astra vs Sol vs Fable 5.1.

Questions people ask

Claude vs ChatGPT vs Gemini for coding: which did best?

Claude did best. All four builds in the Argon review were coding jobs, and Opus 5.5 won each one. Across 21 ranked builds, Opus 5.5 won 16, per the best AI for coding.

Gemini 4 Argon vs Opus 5.5 vs GPT-6 Astra: which won?

Opus 5.5 won all four builds, GPT-6 Astra placed above Gemini 4 Argon in all four, and Argon placed last each time. Argon was the cheapest: $39.54 for the four builds against $189.04 for Opus 5.5 and $409.96 for Astra.

Claude vs ChatGPT vs Gemini price: what does each cost?

All three have a $0 plan. The plan with each top model costs $20 a month for Claude Pro billed monthly, $20 for ChatGPT Plus and $19.99 for Google AI Pro. The biggest personal plans are Claude Ultra at $200, ChatGPT Pro from $100 and Google AI Ultra at up to $199.99, per each company's pricing page (checked November 5, 2026).

Is Claude free?

Yes, Claude has a free plan. Claude's Free plan costs $0 with limited usage every five hours and every week, and it includes Sonnet and Haiku. Opus and Claude Code start on Pro.

Is Gemini 4 Argon free?

No, Argon needs a paid plan. In the Gemini app it needs Google AI Pro ($19.99 a month) or Ultra, per Google's plans page, and Google's API pricing page says Argon is not available on the free tier.

Is GPT-6 Astra free?

No, Astra needs a paid plan. ChatGPT's pricing page lists Astra on Plus ($20 a month) and Pro, and not on Free or Go.

Which is better for writing or research?

Pat's tests don't answer that. He compared the models on four build jobs and didn't test writing, research or everyday questions.

What are some alternatives to Gemini?

Claude and ChatGPT are the two compared here, and both placed above Gemini 4 Argon in Pat's builds. Grok, Meta AI, Copilot and Perplexity are covered, with prices, in ChatGPT alternatives.

How do I switch from one to another?

Cancel the old plan in the place you paid for it, then check that company's refund rules, which depend on where you paid and where you live. The steps for ChatGPT, Claude and Google AI Pro are in how to cancel your ChatGPT subscription.

Update November 5, 2026

November 5, 2026: First published, with results from Pat's Gemini 4 Argon review (October 14, 2026) and plan prices checked on Anthropic's, OpenAI's, Google's and Microsoft's pages that day. Argon's API sale price ends on November 30, so this page will be updated after that.

Sources

• AI for Mortals: Gemini 4 Argon review, test results and costs
• AI for Mortals: Gemini 4 Argon vs GPT-6 Astra on the same 4 builds
• AI for Mortals: The best AI for coding, 21 same-prompt builds ranked
• Anthropic: Claude pricing (checked November 5, 2026)
• OpenAI: ChatGPT pricing (checked November 5, 2026)
• Google: Gemini subscription plans (checked November 5, 2026)
• Google AI for Developers: Gemini API pricing (checked November 5, 2026)
• Google: Introducing Gemini 4 Argon
• Google: Gemini 4 Argon benchmarks
• Google: Gemini 4 Argon in the Gemini app
• Microsoft: Copilot plans for individuals (checked November 5, 2026)
• Microsoft: Compare Microsoft 365 plans (checked November 5, 2026)