Six months ago, picking between ChatGPT and Claude meant choosing between two subscriptions and a handful of models. That's over. Both companies have quietly turned their pricing pages into puzzles — new tiers, renamed models, and a pile of fine print that changes what you actually get for your money depending on which button you click.
OpenAI rolled out its GPT-5.6 family in July, and Anthropic followed with a full repricing of the Claude 5 lineup. The result is that "ChatGPT vs Claude" isn't one question anymore — it's three: which one wins at the free tier, which one wins in the middle where most people actually live, and which one wins at the flagship level where the real capability gap shows up.
Here's how the tiers actually stack up, and where your money goes furthest.
What changed with GPT-5.6 and Claude 5?
Start with ChatGPT. ChatGPT now runs on a seven tier ladder, from a free plan up to a $200 per month Pro plan, with a $20 Plus plan sitting where most people should land, and Plus is the first tier that unlocks the GPT-5.6 flagship model, the ChatGPT Work agent, Sites, and full Codex access. That's a big deal, because the GPT-5.6 family landed in ChatGPT on July 9, with three models carrying names instead of numbers. Those names are Sol, Terra, and Luna — Plus adds GPT-5.6 Sol, Terra, and Luna, expanded deep research, and expanded Codex usage, while on Go, deep research stays limited and the 5.6 reasoning models are not listed at all.
You can check the current lineup yourself on ChatGPT's official pricing page, which OpenAI updates directly.
On the Anthropic side, the reshuffle is arguably bigger. Claude Opus 5 released July 24, 2026, offering near-Fable 5 intelligence at half the price, while Claude Fable 5 is Anthropic's generally available Mythos-class model. And there's a third name floating around that hasn't fully surfaced yet: Mythos 5, which — per the video's framing — stays invite-only for now. Anthropic's actual pricing and model documentation lives at Claude's official pricing docs and claude.com/pricing, both worth bookmarking since these numbers move often.
How do the free tiers compare?
This is where the gap has actually narrowed the most. Anthropic made a surprisingly aggressive move here: the Free tier costs nothing and, since June 30, 2026, defaults to Claude Sonnet 5 — the same mid-tier model that runs on Pro. That's notable because free users aren't stuck with a stripped-down model anymore; they're using the same brain as paying Pro subscribers, just with tighter usage caps. And Anthropic isn't shy about keeping those caps vague — usage caps are described as a rolling budget rather than a message count, and Anthropic does not publish an exact figure for any tier. You get a warning as you approach the limit, then a message naming the reset time.
ChatGPT's free tier tells a different story. OpenAI lists Free's total GPT Instant context window at 27K tokens, roughly 12 pages of input text by OpenAI's own approximation, and ads may appear in the Free tier in certain countries, with Free being appropriate for occasional casual use and testing the interface before committing to a paid plan. And the ads aren't hypothetical anymore — as of February 9, 2026, the Free tier in the US includes advertising, and while OpenAI says ads are clearly labelled and don't influence ChatGPT's answers, they're present.
If you're just testing the waters, Claude's free tier currently offers more model muscle for the same price of zero dollars.
Which mid-tier plan is worth your $20?
This is the tier where most people actually spend their money, and it's also the tier that's changed the least in raw price while changing the most in what you get.
For ChatGPT, if ChatGPT is part of how you work, Plus at $20 is the answer and has been for three years — it's the first tier that gets you GPT-5.6 Sol, the Work agent, Codex, and Sites, and the feature gap between it and Go is far larger than the $12 price difference suggests. In other words, don't assume Go at $8 is "basically Plus but cheaper." Go doesn't include advanced reasoning models, Sora, Codex, Agent Mode, Deep Research, or Tasks — and those aren't nice-to-haves. They're the features that make Plus worth using for professional work.
For Claude, Pro sits at the same $20 price point and has held steady. Pro is $20/month, or $17/month billed annually, and it runs Sonnet 5 by default with access to more Claude models, more usage than Free, Research mode, unlimited Projects, voice mode, and Microsoft 365 integration. Anthropic's Sonnet 5 has also picked up real momentum on the API side, which trickles down to how snappy and capable the consumer product feels — Sonnet 5 currently runs at $2.00 input / $10.00 output per million tokens through August 31, offering 1M context, up to 128K output, and strong performance for everyday coding and production workflows.
Practical takeaway: if your work leans toward writing, structured reasoning, or long documents, Sonnet 5 on Claude Pro is hard to beat at $20. If your work leans toward agentic tasks, image generation, or building things with Codex, Plus earns its price.
What do you actually get at the flagship level?
Here's where the real capability difference shows up — and where the price gap has also gotten more interesting.
On ChatGPT's side, OpenAI split what used to be one brutal jump into two steps. OpenAI now offers two Pro options instead of one gut-wrenching jump to $200 — ChatGPT Pro starts at $100/month and goes to $200/month, splitting into two distinct tiers, with Pro $100 launched April 9, 2026, deliberately positioned against Anthropic's Claude Max at the same price, including GPT-5.5, GPT-5.5 Pro, o1 Pro mode, and 5x Plus usage limits. That $100 tier is genuinely useful if you've outgrown Plus's limits but don't need the full 20x usage multiplier of the $200 plan.
Claude's flagship options run in parallel. Paid consumer plans are Pro at $20/month, Max 5x at $100/month, and Max 20x at $200/month. And the model doing the heavy lifting at the top is Opus 5, which quietly became the better value pick over the more expensive Fable 5 for most agentic and coding work. Claude Opus 5 released July 24, 2026 — near-Fable 5 intelligence at half the price. On raw benchmarks, that shows up directly: Opus 5 edges Fable 5 on SWE-bench Pro, meaning the "flagship" isn't automatically the smartest option anymore — it's Anthropic's premium option, priced accordingly at roughly double Opus 5's API rate.
If you're deciding between the two $100 flagship-adjacent tiers, run the same real task on both before committing. A free multi-model testing ground like aimaster.me lets you compare outputs side by side without paying for two subscriptions just to find out which one handles your specific workload better.
Where does ChatGPT still have the edge?
There's one category where switching to Claude doesn't make sense, no matter how good Opus 5's benchmark scores look: visual output. If your daily work involves generating images, storyboards, mockups, or anything that needs to look like something rather than read like something, ChatGPT's image generation and multimodal tooling remain the more mature product. Claude simply isn't built around that use case the same way, and no amount of reasoning-benchmark superiority changes that gap.
This matters for a specific but sizable group: marketers, social media managers, and anyone producing visual assets on a deadline. If that's your daily output, ChatGPT Plus stays the single most useful seat you can buy, and pairing it with a dedicated design tool covers the gap Claude can't close.
For that visual layer, a tool like Venngage's AI Flyer Generator turns a text description into a finished flyer or visual asset in under a minute — genuinely useful when you need something that looks professional without opening a design program. Venngage is currently running 50% off annual Premium and Business plans with the code AIMASTER50-9.
How do you actually pick between them?
Skip the "which AI is smarter" debate — it's the wrong question. Ask instead:
- What's your actual monthly output? If you're hitting usage caps on Free within days, that's your signal to upgrade, not a hypothetical.
- Does your work depend on visuals? If yes, ChatGPT Plus. If no, that constraint disappears and Claude's writing and reasoning strength becomes the deciding factor.
- Are you doing agentic or coding work? Opus 5's SWE-bench Pro numbers and Claude Code's growth (Claude Code alone has become a massive revenue driver for Anthropic) suggest Claude has quietly become the stronger developer tool.
- Do you need document-heavy context? Claude's 1M token context window across current-generation models handles long documents without the long-context surcharges some competitors charge.
Both companies will keep repricing and renaming things every few months — that's the nature of this market right now. The smart move isn't picking a permanent favorite. It's checking the official pricing pages before you commit, testing your actual workload on both, and upgrading only when a real limit stops your real work — not when a marketing email tells you to.
The AI you should pay for isn't the one that wins the benchmark headlines. It's the one that's still doing your actual job three months from now, without you thinking about it.