OpenAI vs Anthropic (2026): models, pricing, funding and safety compared
OpenAI vs Anthropic in November 2026: ChatGPT wins on free use and breadth, Claude on writing and long context. Plans, API prices, funding and safety compared.
Written byRahis SaifiCo-founder and news editorWhy you can trust The AI Future
Our writers research every product from its official pages, independent tests and user reports, link each claim to its source and correct mistakes openly. Unless an article says otherwise, we haven't tested the product hands-on ourselves. Companies can't pay for coverage or see reviews before they're published.
Read our editorial policy
OpenAI vs Anthropic comes down to breadth against depth in November 2026. OpenAI's ChatGPT has the bigger free plan, the cheaper $8 entry tier, image generation and about 1.2 billion weekly users. Anthropic's Claude wins most published head-to-head tests on writing and reasoning, gives every plan a 1M-token context window, and has just filed for an IPO at a higher private valuation than its older rival.
Both charge $20 a month for the plan most people buy. So the real question is what you do all day.
Nobody at The AI Future ran hands-on tests for this piece. I compared the two companies using their own plan pages, model announcements and API price lists, plus published side-by-side tests from Tom's Guide and DataCamp, financial reporting and user feedback, all as of early November 2026. When someone else did the testing, I name them.
OpenAI vs Anthropic: quick verdict
- For free, everyday use, OpenAI wins. ChatGPT Free offers unlimited text chats with GPT-6 (subject to abuse guardrails), while Claude's free plan has tight five-hour and weekly limits.
- For writing, analysis and long documents, Anthropic wins. Claude won four of five everyday prompts in a Tom's Guide test, and every Claude plan gets up to a 1M-token context window.
- For images, voice with video and an all-in-one app, OpenAI wins. Claude's plan table lists no native image generator.
- For developers, it's close to a tie. GPT-6.1 Sol and Claude Sonnet 5.5 cost the same $2 and $10 per million tokens, and DataCamp's coding test split the decision.
- For a cautious, published safety process, I'd give Anthropic a slight edge, though both labs now publish system cards and external testing.
Our take: I think Claude is the sharper tool for people who write, analyse or code for a living. ChatGPT is probably the better default for everyone else, especially anyone who won't pay.
Side-by-side comparison table
| OpenAI | Anthropic | |
|---|---|---|
| Main consumer app | ChatGPT | Claude |
| Founded | 2015, as a nonprofit lab | 2021, by Dario Amodei, Daniela Amodei and other former OpenAI researchers |
| Top models | GPT-6 Astra, GPT-6.1 Sol, GPT-6 Sol, GPT-6 Luna | Claude Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 5.5 |
| Free-plan model | GPT-6 Luna | Sonnet and Haiku |
| Cheapest paid plan | Go, $8/month (may include ads) | Pro, $20/month |
| Top individual plan | Pro 500, $500/month | Ultra, $200/month |
| Coding agent | Codex (limited on Free) | Claude Code (paid plans only) |
| Image generation | Yes | Not listed |
| Context window | 27K on Free (Instant); larger on paid plans | Up to 1M on all plans |
| Latest private valuation | $852 billion (March 2026) | $965 billion (Series H) |
| Run-rate revenue | About $50 billion net (end of September) | More than $47 billion |
| IPO status | Filed confidentially in June; executives point to 2027 | S-1 filed with the SEC |
Sources for the table: ChatGPT plans, Claude plans, InvestmentNews and Luminix.
Features compared
Models
OpenAI split its GPT-6 family by size and price. GPT-6 Astra is the heavyweight. GPT-6 Sol powers ChatGPT for paid users and GPT-6 Luna runs it for Free and Go users, going by OpenAI's October 7 launch post. A week after GPT-6 Sol, OpenAI shipped GPT-6.1 Sol at DevDay on September 29, pitching it as "near-Astra intelligence for a fifth of the price," DataCamp reports.
Anthropic's newest flagship is Claude Opus 5.5. In its launch post, the company says Opus 5.5 performs at the level of Claude Fable 5.1, its top model, on most work and costs 40% less to run than Opus 5. Sonnet 5.5 and Haiku 5.5 fill out the cheaper end. (The naming on both sides has got a little silly, for what it's worth.)
On the one shared independent yardstick DataCamp could find, the Artificial Analysis Intelligence Index, Sonnet 5.5 scores 56 and GPT-6.1 Sol scores 51.8.
Apps and everyday tools
ChatGPT's big October change was Intelligent UI, which lets answers include charts, buttons, forms and small interactive tools. The plan table also lists deep research, image creation, voice, Sites, scheduled tasks, custom GPTs, ChatGPT Work and, on Pro, an always-on agent called Dot.
Claude goes after work output. Its plan table includes Artifacts, Claude Docs, Slides and Design on every tier, with Claude Code, Claude in Chrome, Research and Claude for Microsoft 365 starting at Pro. It doesn't list an image generator, and that's a real gap if you make visuals.
Context is the quiet split. ChatGPT Free's Instant model gets a 27K-token window, roughly 12 pages of input. Claude offers up to 1M tokens on every plan, free included.
Privacy defaults
Both companies train on consumer chats unless you opt out. That's the same on ChatGPT Free and on every individual Claude plan, so neither has a clear edge here.
Quality of output: same prompts, side by side
We didn't run our own prompts for this comparison. Instead, here are three of the prompts Amanda Caswell, AI editor at Tom's Guide, gave both ChatGPT-6 and Claude Opus 5.5, with what each model produced in her test.
Prompt 1: a $150 birthday party for 10 kids
The brief included two vegetarians, a peanut allergy and a rain backup. ChatGPT budgeted $36 for three delivery cheese pizzas and added a price buffer. Claude flagged cross-contamination risks with bakery cakes and boxed mixes, then used tote bags and fabric markers as both a craft and the party favour. Caswell gave it to Claude, calling its plan "turnkey" and "host-ready."
Prompt 2: two holiday rentals, $1,850 vs $1,500
Rental B was cheaper but needed a 25-minute drive and $35 a day for parking. ChatGPT raised the best single point: a 10-minute walk with three kids and beach gear isn't always easier than driving. Claude did the full maths, found the real saving was only about $55 to $65 for the week, and warned that cleaning fees and taxes could flip the result. Claude won "by a hair."
Prompt 3: a system for a disorganised parent
Here ChatGPT won. Caswell said it built a decision filter that protected "an exhausted parent's mental bandwidth," while Claude gave a good but more standard home routine.
Claude took four of the five rounds overall. Her summary of the difference was short: "Claude tends to keep digging." She also found that ChatGPT gets to the point faster, which helped in that final test.
Coding tells a different story. In DataCamp's hands-on coding test, Matt Crabtree found that "Sol built the more careful app, but Sonnet got there faster and in fewer turns." GPT-6.1 Sol scored 5.0 on his rubric to Sonnet 5.5's 4.7. One test from each outlet is a small sample, so I wouldn't treat either result as settled.
Pricing compared
The consumer plans line up neatly at $20, then split apart. OpenAI has a cheaper rung below it and a pricier one above it.
| Plan | OpenAI (ChatGPT) | Anthropic (Claude) |
|---|---|---|
| Free | $0, GPT-6 Luna, unlimited text chats | $0, Sonnet and Haiku, limited |
| Entry | Go, $8/month | None |
| Standard | Plus, $20/month | Pro, $20/month (cheaper billed annually) |
| Power user | Pro 100, $100/month; Pro 200, $200/month | Max, $100/month (5x Pro usage) |
| Top tier | Pro 500, $500/month (adds Ultrafast models) | Ultra, $200/month (10x Pro weekly usage) |
| Teams | Business, $20/user/month annually or $25 monthly | Team Standard, $25/seat/month |
Prices checked in early November 2026 from the ChatGPT Plus and ChatGPT Pro help pages, Anthropic's Pro and Max articles, The Next Web for Ultra, and eWeek's pricing cheat sheet for Go and the team plans.
Ultra is new. Anthropic renamed its $200 Max 20x plan to Ultra and dropped the five-hour limit for verified accounts, The Next Web reported. The catch: the plan page now lists ten times Pro usage a week, and the old name promised 20.
Usage limits matter more than list price on Claude. TechRadar quoted AI educator Kyle Balmer saying that on the $20 plan, "realistically Sonnet is the only model most people can actually use." I suspect that's the single biggest reason people bounce back to ChatGPT.
API prices
| Model | Input per 1M tokens | Output per 1M tokens |
|---|---|---|
| GPT-6 Astra | $10.00 | $50.00 |
| Claude Opus 5.5 | $4.00 | $20.00 |
| GPT-6.1 Sol | $2.00 | $10.00 |
| Claude Sonnet 5.5 | $2.00 | $10.00 |
| GPT-6.1 Terra | $1.00 | $5.00 |
| GPT-6 Luna | $0.10 | $0.50 |
Standard rates are from OpenAI's API price list and Anthropic's Opus 5.5 post. The fine print differs: GPT-6.1 Sol's cached input is $0.10 per million against Sonnet's $0.20, but Sol charges double for input above 272K tokens, DataCamp notes, while Sonnet keeps one rate across its 1M window. OpenAI also goes much cheaper at the bottom, and Luna's $0.10 input price is hard to beat for high-volume work. If tokens are new to you, our guide to what a token is in AI explains the billing unit.
Funding, revenue and business model
Anthropic is now the more valuable private company on paper. It closed a $65 billion Series H at a $965 billion post-money valuation, led by Altimeter, Dragoneer, Greenoaks and Sequoia, InvestmentNews reported. The company says run-rate revenue passed $47 billion, and its newsroom confirms it has filed an S-1 for a proposed IPO.
OpenAI raised $122 billion at an $852 billion post-money valuation in March, with Amazon committing up to $50 billion and Nvidia and SoftBank $30 billion each, according to The Next Web. Bloomberg has reported it wants about $1.4 trillion in a new round, which Luminix notes isn't signed.
Comparing their revenue is trickier than the headlines suggest. Luminix explains that Anthropic books the full payment on sales through cloud partners such as Amazon and Google, while OpenAI books only its own share of certain Azure resales. On OpenAI's method, its run rate was about $50 billion at the end of September; counted Anthropic's way, investors put it at $68 to $70 billion. So the two are probably closer in size than the valuations imply.
Neither is profitable on the numbers I've seen. OpenAI's audited 2025 revenue was $13.07 billion against a $20.92 billion operating loss, Luminix reports.
Safety approach compared
Anthropic builds its pitch around safety, and its paperwork is the more detailed of the two. Its Responsible Scaling Policy has gone through nine versions since September 2023, with version 3.4 effective July 8, 2026, plus regular Risk Reports (the latest from August 2026). Opus 5.5 was tested before release by outside evaluators including METR, and Anthropic says it scored best of any model on its automated behavioural audit. Anthropic also kept its limits on mass surveillance and autonomous weapons when OpenAI signed a Pentagon deal, TechRadar reported.
OpenAI uses its Preparedness Framework, last overhauled in April 2025, which tracks risks that are "plausible, measurable, severe, net new, and instantaneous or irremediable." For GPT-6, OpenAI says it ran alignment evaluations throughout training and saw stronger resistance to jailbreaks that adapt over several turns. It rates GPT-6.1 Sol as Critical in cybersecurity, DataCamp notes, and sells separate gated cyber and life-sciences models on its API.
Both labs have their blemishes. A Claude Code data leak earlier this year raised data-handling questions, per InvestmentNews, and a June lawsuit claimed Anthropic oversold Max usage. OpenAI, for its part, shut down its Sora app after engagement fell and a Disney licensing deal collapsed. In my view, Anthropic is more open about its process; whether that makes its models safer in practice is something outsiders can't fully check.
Reviewers and users on each
On Claude, Tom's Guide's Caswell was clear: "Claude tends to keep digging." On ChatGPT, she found "ChatGPT always seems to be the less 'fluffy' of the two models, with a response that gets to the point faster."
Users are harder on both. One poster on OpenAI's developer forum wrote that since GPT-6 became the default, ChatGPT "has felt even less intelligent than the models we had back in 2023," and called the new interface elements "flashy, unnecessary." Another reply in the same thread still called GPT "the best LLM out there overall."
Claude's main complaint is limits. "For new users, it's a shock," Balmer told TechRadar, pointing to how fast Opus eats the allowance.
Which one should you choose?
Pick OpenAI's ChatGPT if you want one app that does most things, won't pay or want to pay just $8, or need images and voice.
Pick Anthropic's Claude if your day is long documents, careful writing or coding, and you'll pay $20 or more. Developers on AWS or Google Cloud get Sonnet 5.5 there directly, which DataCamp couldn't confirm for GPT-6.1 Sol.
Honestly, I'd guess plenty of heavy users will end up paying for both, and at $40 a month that's not unreasonable.
For a broader view, see our Claude AI vs ChatGPT comparison, our OpenAI review and Claude AI review, or our list of AI tools like ChatGPT. If Google's in the mix, Google AI vs ChatGPT covers that matchup.
FAQ
Is Anthropic better than OpenAI?
It depends on the job. Claude, from Anthropic, won four of five everyday prompts in a Tom's Guide test and has a 1M-token context window on every plan. ChatGPT, from OpenAI, has the more generous free plan, an $8 tier and image generation. For coding, DataCamp's test came out close to even.
Is Anthropic owned by OpenAI?
No. They're rival companies. Former OpenAI researchers, including Dario and Daniela Amodei, founded Anthropic in 2021, and its latest round was led by Altimeter, Dragoneer, Greenoaks and Sequoia.
Which is cheaper, ChatGPT or Claude?
ChatGPT is cheaper to start, with an $8 Go plan and an unlimited free text chat. Both main plans cost $20 a month. At the top, Claude's Ultra costs $200 while ChatGPT Pro runs to $500. On the API, GPT-6.1 Sol and Claude Sonnet 5.5 share the same $2 and $10 per million tokens.
Which company is worth more, OpenAI or Anthropic?
On their last closed rounds, Anthropic, at a $965 billion post-money valuation, against OpenAI's $852 billion from March 2026. OpenAI is reportedly seeking about $1.4 trillion in a new round, and secondary trading put it near $987 billion in October, so the order could flip soon.
Which is safer, OpenAI or Anthropic?
Both publish safety frameworks and system cards. Anthropic's Responsible Scaling Policy is updated often and paired with public Risk Reports, and Opus 5.5 had outside testing by METR. OpenAI uses its Preparedness Framework and adversarial testing for GPT-6. Outsiders can't fully verify either company's claims.
