# Claude vs Gemini: Opus 5.5 vs Gemini 4 Argon on the Same Builds, and Which to Use for Each Job > Pat Simmons gave Claude's Opus 5.5 and Google's Gemini 4 Argon the same four builds and ranked them blind. Opus placed first every time and Argon last, Argon cost far less to run, and each starts on a plan of about $20 a month. By Pat Simmons · November 6, 2026 · 10 min read Canonical: https://www.aiformortals.co/blog/claude-vs-gemini Use Claude for building things with AI, like apps, games, 3D scenes and launch videos, and Gemini when the cost of each job comes first or you already work in Gmail and Google Docs. In Pat Simmons's blind test of Claude's Opus 5.5 against Google's Gemini 4 Argon, Opus placed first and Argon last on every build, while Argon cost a fraction as much to run. Both top models start on a plan of about $20 a month: Claude Pro at $20 and Google AI Pro at $19.99. Last checked November 6, 2026: plan prices and which plan includes each top model, on Claude's pricing page and Google's Gemini plans page, plus per-token prices for Opus 5.5 on Anthropic's API pricing page and for Argon on Google's API pricing page. Google's sale price for Argon ends on November 30, 2026, and this page will be checked again after that date. ## Claude vs Gemini: our verdict, by job For build work, Claude placed higher every time. Pat gave Claude's Opus 5.5 and Fable 5.1, ChatGPT's GPT-6 Astra and Gemini 4 Argon the same prompts, ranked the results blind, and only then saw which model made which. Opus 5.5 came first and Argon last on all four builds. ModelApp3D bottleLaunch videoCar soccer gameWebsite rebuild Opus 5.5Claude1st1st1st1st Fable 5.1Claude2nd3rd2nd or 3rd2nd or 3rd GPT-6 AstraChatGPT3rd2nd2nd or 3rd2nd or 3rd Gemini 4 ArgonGemini4th4th4th4th Ranks are Pat's blind picks out of four models. The bottle and launch video ranks are from the reveals in the video above. The game and website ranks are from Pat's full Gemini 4 Argon review, where Fable 5.1 and Astra split second and third. The next table shows what each build cost and how long it took. These are API-equivalent costs: the tokens each build used (tokens are the small chunks of text a model reads and writes), priced at the model's published rate. Build Opus 5.5: rank, cost, time Gemini 4 Argon: rank, cost, time 3D bottle1st, $29.95, 114 min4th, $6.85, 36 min Launch video1st, cost in the full review, 318 min4th, $10.14, 80 min Car soccer game1st, $46.24, 155 min4th, $10.59, 54 min Website rebuild1st, $67.87, 129 min4th, $11.95, 77 min The bottle figures are on screen at the reveal shown below. The other figures come from the cost tables in the full review. By job, that works out like this: - Apps, games, 3D and video builds: Claude (Opus 5.5). - The lowest cost per job: Gemini (Argon), at least until November 30. - Work that lives in Gmail and Google Docs: Gemini, which is built into both on AI Pro. - Writing and research: untested by Pat; see the feature comparisons below. If you're weighing ChatGPT as well, GPT-6 Astra placed between the two, above Argon on every build. The three-way results are in Claude vs ChatGPT vs Gemini on the same builds, and the two-way ones in Gemini vs ChatGPT: Argon vs GPT-6 Astra. ## Is Claude better than Gemini? In Pat's builds, yes. Both Claude models placed above Gemini 4 Argon on all four jobs, and Opus 5.5 won each one. After the launch video round, Pat said, "Opus 5.5 is on a generational run right now with these model tests." The first build was a 3D product. Each model had to build a photoreal, interactive cold brew bottle for a made-up coffee brand, Clanker Coffee, with a glass of ice and a pour you trigger with a click. Argon's bottle came out looking illustrated, and at the reveal Pat said, "Opus just crushes 3D work." _The blind reveal for the 3D bottle: Opus 5.5 (column D) ranked #1, Fable 5.1 #2 and GPT-6 Astra #3, with Gemini 4 Argon's column left fourth. Each card shows the build's cost, input and output tokens and time. From Pat's review at 13:42. Open the builds: Opus 5.5, Gemini 4 Argon._ Google's launch demos leaned on detailed 3D scenes built in Blender, and Pat said Argon's attempt "wasn't even freaking close to their blog posts." He also called it "still not a bad job overall from Argon," and noted each build was a single attempt. The launch video went the same way. Every model got Pat's launch video skill, GPT Image Gen 2.5 for still images and Blender for 3D. Opus 5.5 placed first, GPT-6 Astra second, Fable 5.1 third and Argon fourth, and Pat said Fable 5.1 and Argon were close. He singled out how the Opus video opens on a "hit enter" idea and cuts to a robot whose eyes light up. _The opening shot of Opus 5.5's launch video, the "hit enter" idea Pat singled out. From Pat's review at 20:37. Watch Opus 5.5's full launch video._ Argon's 3D work was better in its launch video than on the bottle, Pat said, but it had so much text on screen that it never felt like a proper commercial. _Gemini 4 Argon's launch video 29 seconds in, with three product cards full of small text over the 3D scene. From Pat's review at 18:44. Watch Argon's full launch video._ Gemini comes out ahead on price, and on Google's own tests of office and legal work. Google's Argon post says it ranks first on AutomationBench, Zapier's test of business tasks, with 51.3%, and leads on Harvey's Legal Agent Benchmark, a legal research and drafting test. Those are Google's numbers, and Pat didn't test that kind of office work. ## Claude vs Gemini for coding Claude did better at coding in Pat's test. All four builds were coding jobs, where the models wrote the code themselves and, for the 3D work, drove Blender through Python scripts, and Opus 5.5 placed first each time while Argon placed last. For everyday coding, Pat points to one benchmark. In the review he said Terminal-Bench "is usually quite indicative of the day-to-day coding tasks," and that "Opus, being at the top of Terminal Bench, is just the best day-to-day coder on any kind of task you're giving it." Google's own benchmark table shows Argon at 57.4% on Terminal-bench 4.0, against 66.4% for Opus 5.5, and Pat said Argon "really isn't even competing with Opus 5.5" there. The same table puts Argon ahead of the other three models on DeepSWE, a test of long software engineering tasks, with 77.9% (Google's figure). Pat was wary of that one: Muse Wonder scored 79 on it, and in his testing Muse Wonder "is indeed not the best coding agent." To code with Claude, Claude Code comes with Pro and every plan above it, per Claude's pricing page. Google's plans page lists its coding agent, Antigravity, on AI Pro with entry rate limits, and with higher limits on Ultra. Pat found that Argon's usage runs out fast. He ran it on a Google subscription and said, "I was on the $99 a month plan [Google AI Ultra, listed at $99.99], but it only was able to get through about one build, this 3D cold brew bottle before it already hit daily usage limits." He put it at a little faster than Codex or Claude Code. ## Claude vs Gemini for writing Pat hasn't compared Claude and Gemini on writing, so this section has no verdict from use. It covers what each app offers, from the companies' own pages. Claude includes Claude Docs on every plan, free included, plus Projects (up to five on the free plan), per Claude's plan comparison. Its Google Workspace add-on puts Claude in a sidebar in Google Docs, Sheets and Slides, where it reads the file you have open and edits it in place, with every change saved in the version history. More on that in Claude now works inside Google Docs, Sheets and Slides, and to get drafts in your own voice, see how to teach Claude your writing style from your own edits. Gemini is built into Google's apps. Google AI Pro includes Gemini in Gmail, Docs and Vids, and AI Plus includes it in Gmail and Vids, per Google's plans page. In the Gemini app you can preview and edit the documents, PDFs, spreadsheets and decks it makes right in the chat (our story: Gemini can now edit the files it makes), and Gemini in Google Slides can rework a whole deck from one prompt. Google also says Argon "follows style instructions better than any of our previous Gemini models" (Google's Argon announcement). ## Claude vs Gemini for research Both apps can research a question across the web and your own files, and Pat hasn't tested which does it better. Gemini's Deep Research is included on the free plan, per Google's plans page. According to Google's Deep Research help page, Gemini first writes a research plan you can edit, searches Google by default, and can also draw on your Gmail, your Drive or files you upload. A report usually takes 5 to 10 minutes. Claude can search the web on every plan, and every answer that uses web search includes citations. Its deeper Research mode needs Pro or higher. Per Anthropic's help page, Research runs several searches that build on each other, can also search connected apps like Gmail, Google Calendar and Google Docs, and runs on Opus 5.5 whichever model you picked. The same page says it uses up your Claude limits faster than a normal chat. ## Claude Pro vs Gemini Pro: what each costs Claude Pro costs $20 a month and Google AI Pro $19.99, and each is the cheapest plan with that company's top model: Opus on Claude, Gemini 4 Argon on Gemini. Neither top model is on the free plan. Plan level Claude Gemini FreeFree, $0: Sonnet and Haiku, limited use every 5 hours and every weekFree, $0: the Flash-Lite model, Deep Research, 15 GB of storage Under $5No plan at this levelGoogle AI Plus, $4.99: the Flash model, twice the free usage, 400 GB About $20Pro, $20 billed monthly or $17 a month billed yearly: Opus, Claude Code, ResearchGoogle AI Pro, $19.99: Gemini 4 Argon, four times the free usage, Gemini in Gmail and Docs, 5 TB About $100Max, $100: 5 times Pro's usageGoogle AI Ultra, $99.99: 5 times AI Pro's usage About $200Ultra, $200: no 5-hour limit for verified accounts, 10 times Pro's weekly usageGoogle AI Ultra, $199.99: 20 times AI Pro's usage US monthly prices from Claude's pricing page and Google's Gemini plans page, checked November 6, 2026. More on Claude's $0 plan in what Claude's free plan includes. _Claude's plans (left) and Google's Gemini plans (right), captured November 6, 2026. Only Google AI Pro and Ultra carry the "Includes Gemini 4 Argon" label. Sources: Claude pricing, Google's Gemini plans._ For Gemini Ultra vs Claude Max, the prices match: Google AI Ultra starts at $99.99 a month for five times AI Pro's usage, and Claude Max costs $100 for five times Pro's usage. Max also lets you use Fable for up to half your weekly limit and adds $100 a month in API credits, per Claude's pricing page. Paying per use through the API, Google lists Argon at $2 per million input tokens and $10 per million output tokens through November 30, 2026, then $4 and $20 from December 1, on its API pricing page. Anthropic lists Opus 5.5 at $4 and $20 on its pricing page, so Argon costs half as much until the sale ends and the same after. At the sale price, Pat called Argon "by far the cheapest frontier model." ## What are the limits and pros and cons of each? Claude's main limits are its usage caps and the cost of its top model. Every Claude plan caps usage every five hours and every week, except Ultra, which drops the five-hour limit for verified accounts, per Claude's pricing page. The free plan has no Opus, and in Pat's 3D bottle build Opus 5.5 was the most expensive and slowest model, at $29.95 and 114 minutes. Gemini's strengths, from Pat's test and Google's pages: - The lowest cost on every one of Pat's builds - Gemini in Gmail and Docs on AI Pro - 5 TB of storage on AI Pro - Deep Research on the free plan Its weak spots: - Last place in all four of Pat's builds - Argon needs a paid plan - Usage caps every 5 hours, up to a weekly limit - Argon's sale price ends November 30 Google AI Pro and Ultra subscribers get Argon in the Gemini app and in AI Mode, per Google's availability post, and the Gemini app's usage limit refreshes every five hours until you reach a weekly limit, per the footnotes on Google's plans page. For what has changed in each app since, see Claude updates and features and Gemini updates and features. Does Gemini AI spy on you? Gemini saves your chats, and while its Keep Activity setting is on, Google uses them to improve its services, including training its AI models, and human reviewers read some of them, per Google's Gemini Apps Privacy Hub. Reviewed chats are kept for up to three years even if you delete your activity. The rest auto-delete after 18 months unless you switch that to 3 or 36 months. With Keep Activity off, or in a temporary chat, Google doesn't use your future chats to train its models unless you send feedback, but it still keeps them for 72 hours. Claude has a similar switch. On the Free, Pro, Max and Ultra plans, Anthropic uses your chats to improve its models if you allow it in your privacy settings or if a chat is flagged for a safety review, per Anthropic's privacy center. If you allow it, Anthropic may keep de-identified chats for up to five years, and Incognito chats are never used to improve Claude (how long Anthropic stores your data). ## How Pat tested Claude and Gemini Pat Simmons ran the test for his Gemini 4 Argon review, published October 14, 2026, with Claude's Opus 5.5 and Fable 5.1 and ChatGPT's GPT-6 Astra alongside Argon. Every model got the exact same prompt for each build and ran on high effort. He judged each round blind, with the builds labeled A to D, and revealed which model made which only after ranking them. For the 3D work the models ran Blender headless, meaning without its on-screen interface, through Python scripts, and the launch video round added GPT Image Gen 2.5 and his launch video skill. The costs are API-equivalent, worked out from each build's token count. Pat ran Argon on a Google subscription rather than paying per token, and each build was one attempt. Every prompt and every live build is in the full Gemini 4 Argon review. ## Questions people ask Is Claude free? Yes, Claude has a free plan. It costs $0, includes Sonnet and Haiku with limited usage every five hours and every week, and leaves out Opus and Claude Code, which start on Pro at $20 a month (Claude's pricing page). Is Gemini 4 Argon free? No. In the Gemini app it needs Google AI Pro ($19.99 a month) or Ultra, and Google's API pricing page says Argon is not available on the free tier. Which is cheaper, Claude or Gemini? Gemini cost less in Pat's builds: Argon's 3D bottle cost $6.85 against $29.95 for Opus 5.5. The paid plans cost about the same, $19.99 for Google AI Pro and $20 for Claude Pro. Can Claude work inside Google Docs? Yes. Claude for Google Workspace, a beta add-on, opens Claude in a sidebar in Docs, Sheets and Slides and edits the open file. It can't see your other Drive files or your email unless you turn on connectors, per Anthropic's help page. ## Sources - AI for Mortals: Gemini 4 Argon review, test results and costs - AI for Mortals: Claude vs ChatGPT vs Gemini on the same builds - Anthropic: Claude pricing (checked November 6, 2026) - Anthropic: API pricing (checked November 6, 2026) - Google: Gemini subscription plans (checked November 6, 2026) - Google AI for Developers: Gemini API pricing (checked November 6, 2026) - Google: Gemini 4 Argon benchmarks - Google: Introducing Gemini 4 Argon - Google: Gemini 4 Argon in the Gemini app - Google: Gemini Apps Privacy Hub - Google: Use Deep Research in Gemini Apps - Google: Generate, preview and edit files in Gemini - Anthropic: Use Research on Claude - Anthropic: Enable and use web search - Anthropic: Use Claude in Google Docs, Sheets and Slides - Anthropic: Is my data used for model training? - Anthropic: How long do you store my data?