# Claude vs ChatGPT: 9 Head-to-Head Builds, Plans, Prices

> Which one to pay for by job, from Pat Simmons's nine same-prompt builds of Claude and ChatGPT models, plus what Claude Pro and ChatGPT Plus cost.

By Pat Simmons · November 11, 2026 · 10 min read

Canonical: https://www.aiformortals.co/blog/claude-vs-chatgpt

For building apps, sites and games, Claude has the edge right now: in Pat Simmons's latest same-prompt head-to-head, published September 25, 2026, Claude's Opus 5.5 beat ChatGPT's GPT-6 Astra on all three builds, after ChatGPT's models came out ahead in his two rounds before that. Pat still uses Astra day to day for big refactors because it is "methodical", ChatGPT makes images in the chat and Claude doesn't, and he hasn't compared the two for writing or research, so pick by plan for those. Claude Pro and ChatGPT Plus both cost $20 a month.

Last checked November 11, 2026: plan prices and what each plan includes, on Anthropic's Claude pricing page, OpenAI's ChatGPT Plus help page and ChatGPT Pro tiers page; the build results are from Pat's videos of July 11, September 5 and September 25, 2026. Anthropic has since released Fable 5.2, which Pat reviewed in Fable 5.2: NO-HYPE Full Review (& Testing); it isn't part of these head-to-heads.

## Claude vs ChatGPT: the verdict by job

Claude is the pick for one-shot build work in Pat's most recent round, ChatGPT is the pick for making images, and the two are level on price. Claude is made by Anthropic and ChatGPT by OpenAI; both have a free plan and a $20-a-month plan.

- Building apps, sites and games from one prompt: Claude (Opus 5.5)

- Long, careful refactors: GPT-6 Astra, Pat's own day-to-day choice

- Making pictures: ChatGPT

- Writing and research: untested by Pat; compare the features below

- Price: a tie at $20 a month

The winner has changed from round to round. Across 9 builds in three videos, the Claude model won all 3 builds in September 25's round, GPT-6 Astra won all 3 in September 5's round, and GPT-5.6 Sol took July 11's round. Each row below is one build, with Pat's pick as he called it on camera.

Video dateBuild Claude model ChatGPT modelPat's pickClaude timeChatGPT timeClaude costChatGPT cost

Sep 25, 2026Moby Dick scenes drawn only in code, in a risograph print styleOpus 5.5GPT-6 AstraOpus 5.585 min30 min$37.22$8.16
Sep 25, 2026A remake of the Opus 5.5 launch film and web page, with sound made in codeOpus 5.5GPT-6 AstraOpus 5.540 min23 min$11.66$7.96
Sep 25, 2026A Tony Hawk's Pro Skater-style game in a warehouseOpus 5.5GPT-6 AstraOpus 5.5142 min49 min$58.05$44.08
Sep 5, 2026A 3D iPod app loaded with AI-made early-2000s songsFable 5.1GPT-6 AstraGPT-6 AstraAbout 1 hr 12 min to 1 hr 20 min37 min$41 (Claude Code's own readout)About $13
Sep 5, 2026A 3D reconstruction of D-Day you can scrub throughFable 5.1GPT-6 AstraGPT-6 AstraAbout an hour or less10 hr 51 minNot givenNot given
Sep 5, 2026A Call of Duty-style shooter on the Nuketown mapFable 5.1GPT-6 AstraGPT-6 AstraAbout 47 minAbout 47 minNot givenNot given
Jul 11, 2026A pixel-for-pixel clone of the Shots.so mockup appFable 5GPT-5.6 SolGPT-5.6 Sol30 min2 hr 51 min$12Almost $100
Jul 11, 2026An invented viral internet stunt and its drop websiteFable 5GPT-5.6 SolGPT-5.6 SolUnder 20 minAbout 30 min$20About $5
Jul 11, 2026A New York City learning platform from a deliberately rough briefFable 5GPT-5.6 SolSol's design first; Fable's 3D map after one round of feedbackAbout 20 min11 min (first pass)$95$11

Costs are API-equivalent estimates: what each build's tokens (the small chunks of text a model reads and writes) would have cost at the model's published API price, worked out by an agent from the session logs. Pat ran every build on a subscription, so no subscriber paid these amounts, and he says to take the numbers with a grain of salt. The September 25 figures are the ones in the full cost, token and time table for the Opus 5.5 vs GPT-6 Astra builds, which counts active build time only.

## Is Claude better than ChatGPT?

It depends on the month you ask. Claude's Opus 5.5 is ahead as of Pat's September 25, 2026 round, and three weeks earlier ChatGPT's GPT-6 Astra beat Claude's Fable 5.1 on every build.

In July, Pat gave Claude's Fable 5 and ChatGPT's GPT-5.6 Sol the same three builds. He picked Sol on the Shots.so clone and the stunt website, and at the end said, "I really think Sol took the cake here", calling it more thorough, better at following directions and better at design. On the New York City platform he leaned toward Fable's 3D map once both models had one round of feedback.

On September 5, GPT-6 Astra beat Claude's Fable 5.1 on all three builds, and Pat called it "not even close." His conclusion that day: "I think Astra is the best coding model. For the first time, OpenAI has surpassed Anthropic." That was before Opus 5.5 came out.

_GPT-6 Astra's D-Day reconstruction, with 3D troops you can inspect and a timeline you can scrub. Astra worked on it for 10 hours 51 minutes; Pat found no 3D soldiers in Fable 5.1's version. From Pat's GPT-6 Astra vs Fable 5.1 video at 17:35._

On September 25, Pat gave Opus 5.5, Fable 5.1 and GPT-6 Astra the same three builds and picked Opus 5.5 each time: "Opus 5.5 across the board seems to be the clear winner." He preferred Fable 5.1's version to Astra's on all three builds too.

The best version of Claude in Pat's tests is Opus 5.5, which comes with Claude Pro and the bigger plans and isn't on the Free plan. Opus 5.5 also beat GPT-6 Astra and Google's Gemini 4 Argon in a separate blind-ranked test: Claude vs ChatGPT vs Gemini on the same four builds. Both apps change often, and we keep what's new in Claude and what's new in ChatGPT on their own pages.

## Claude vs ChatGPT for coding

For one-shot builds Claude's Opus 5.5 is the better pick in Pat's latest round, and for everyday engineering work Pat himself still reaches for ChatGPT's GPT-6 Astra. In each test the Claude models ran in Claude Code and the ChatGPT models in Codex, OpenAI's coding agent, both on high effort.

Pat's wrap-up on September 25 put it this way: "I'm still partial to Astra in my day-to-day", naming "big refactors, large like multi-step site architecture" and 3D work, because "I really like how methodical Astra is." He added that the builds are "not totally indicative of the more realistic everyday tasks." On Fable 5.1 he said "it's way faster, but then it misses all of these things that I ask it to do", and that he had yet to see that from Opus 5.5.

The skate game shows the gap. Opus 5.5 took 142 minutes against Astra's 49, and in Pat's play-through its version had the most realistic skater, light through the warehouse windows, controller support and grinds you have to balance.

_Opus 5.5_

_GPT-6 Astra_

Both skate games are live: click either start screen to play it.

Usage limits differed too. Pat hit his Claude limits during the Moby Dick build and waited about 30 minutes for them to reset, on the 20x plans for both Claude and ChatGPT. Astra's nearly 11-hour D-Day build didn't hit a limit on his $200-a-month ChatGPT Pro plan. Claude Code comes with every paid Claude plan, and if you're choosing between coding tools, see Cursor vs Claude Code, by job.

## Claude vs ChatGPT for writing

Pat hasn't tested Claude and ChatGPT against each other for writing, so there is no winner from use here, only what each company says its plans include. On the free plans, ChatGPT allows a limited number of messages with uploads, such as an old resume to rework, and Claude lists file creation and Docs.

For keeping your voice and details between chats, both have memory. Claude's Free plan lists "Memory across conversations", and OpenAI's memory help page says ChatGPT can draw on past chats, saved memories and custom instructions. Both have projects for grouping related chats and files: up to 5 on Claude's Free plan and more on Pro, and on every ChatGPT plan, per OpenAI's plan comparison.

For long documents, Anthropic's context window page says Opus 5.5 has a 1 million token context window (how much of a conversation it can keep in view) in chat on paid plans, and puts 200,000 tokens at about 500 pages of text. OpenAI lists ChatGPT Plus at about 320 pages of text for its reasoning models and Pro at about 680 pages.

## Claude vs ChatGPT for research

Both have a research mode that runs many web searches and writes up what it found; Claude's is on paid plans only, and ChatGPT's comes in a limited form on the Free plan. Pat hasn't compared them.

Anthropic's Research help page says Claude runs multiple searches that build on each other, can also search connected tools like Gmail, Google Calendar and Google Docs, and runs on Opus 5.5 whichever model you picked. It uses the same usage limits as ordinary chats and can use them up faster.

OpenAI's deep research page says ChatGPT proposes a research plan you can edit before it starts and ends with a report that has citations or source links. OpenAI's plans list deep research as limited on Free, expanded on Plus and at its maximum on Pro.

## Is Claude or ChatGPT better for image generation?

ChatGPT, because Claude doesn't make pictures. Anthropic's help page on images says Claude "doesn't generate photos or illustrations the way image-generation tools do"; it builds diagrams, charts and interactive visuals in the chat instead, in beta on web and desktop.

ChatGPT creates images from a description or from a photo you upload, per OpenAI's Go plan page. Image creation is limited on the Free plan, and Plus and Pro add image generation with Thinking. Pat's video ChatGPT Images 2.0 is WILD - 50+ Use Cases shows it in action.

## Claude Pro vs ChatGPT Plus: what each costs

Claude Pro and ChatGPT Plus both cost $20 a month, and only Claude offers a yearly discount, at $17 a month paid as $200 up front.

PlanMonthly priceYearly billingTop modelsCoding agentImagesIn-depth research

 Claude Pro$20$17 a month, $200 up frontOpus, Sonnet and Haiku; Fable paid with usage creditsClaude Code includedNo photos or illustrationsResearch included
 ChatGPT Plus$20None listed; billed monthly"Advanced reasoning models with GPT-6""Expanded Codex usage""More complex and accurate image creation""Expanded deep research"

Prices are in US dollars, from Anthropic's Claude pricing page and OpenAI's Plus help page, checked November 11, 2026; the quoted Plus features are from OpenAI's plan page. Both companies also have a Free plan: Claude's includes Sonnet and Haiku but not Opus, and ChatGPT's lists "Unlimited text chats with GPT-6" with limited uploads, images and deep research.

Above those, Claude Max costs $100 a month for 5 times Pro's usage and Claude Ultra costs $200 for 10 times Pro's weekly usage. ChatGPT Pro comes in three tiers, Pro 100, Pro 200 and Pro 500, at $100, $200 and $500 a month, so Claude Max and the cheapest ChatGPT Pro tier cost the same. OpenAI's usage limits page says that from December, Plus, Pro and Business will share one allowance across ChatGPT and Codex.

For the full list of OpenAI's plans and the limits on each, see what ChatGPT Plus gets you and its limits.

## How Pat tested Claude and ChatGPT

Each round gave a Claude model and a ChatGPT model the exact same prompt, once, with no revisions, except the July New York City build, where Pat gave both models one round of feedback and judged both versions. Each build was deployed live to the internet. All models ran on high effort, the Claude models in Claude Code and the ChatGPT models in Codex. Pat judged the results on camera by playing with each build; these were not blind rankings.

- September 25, 2026: Opus 5.5 and Fable 5.1 vs GPT-6 Astra

- September 5, 2026: Fable 5.1 vs GPT-6 Astra

- July 11, 2026: Fable 5 vs GPT-5.6 Sol

Pat ran the builds on subscription plans, which don't show what a session costs, so he had an agent estimate each build's API-equivalent cost from the session logs and said the figures are rough. In July he was on a $100-a-month ChatGPT plan and a $200-a-month Claude plan; in September, the 20x plans on both. The tests cover big creative builds only, and Pat said they are "not totally indicative" of everyday tasks.

## Questions people ask

Is Claude free?
Yes. Claude's Free plan costs $0 and includes the Sonnet and Haiku models, web search, file creation and memory. Opus, Claude Code and Research need Pro or above.

Is ChatGPT free?
Yes. ChatGPT's Free plan has unlimited text chats with GPT-6 and limited uploads, image creation, voice chats, deep research and Codex.

Is Claude Max or ChatGPT Pro better value?
Claude Max and ChatGPT's Pro 100 tier both cost $100 a month. Anthropic says Max gives 5 times Claude Pro's usage, while OpenAI's Pro tiers page says only that Pro 200 includes more usage than Pro 100 and Pro 500 the most. Pat hasn't compared the two plans; his September builds ran on the 20x plans of both.

## Sources

- Claude plans and pricing (Anthropic)

- ChatGPT plans and pricing (OpenAI)

- What is ChatGPT Plus? (OpenAI Help Center)

- About ChatGPT Pro tiers (OpenAI Help Center)

- I Made GPT-6 Astra & Fable 5.1 Build the Same App (RAW RESULTS), Pat Simmons

- I Made GPT 5.6 and Fable 5 Build the Same App (RAW RESULTS), Pat Simmons

- Fable 5.2: NO-HYPE Full Review (& Testing), Pat Simmons

- ChatGPT Images 2.0 is WILD - 50+ Use Cases, Pat Simmons