AI Price Tracker 2026: Every Tool's Price + Real Free-Tier Limits (Live)

The live instrument of The Nesyona LLM Audit, the independent standard behind every Nesyona verdict. See the per-task math in The Shrinkflation Index.

Pricing pages for AI tools are designed to look stable. They mostly are -- the same "$20/month" headline has sat on Claude and ChatGPT since 2023. But three things happened in early 2026 that changed the real cost without touching the sticker. This page tracks all of it: the live price table, the real free-tier limits (not the marketing copy), and a dated log of every material change we have observed.

The data layer underneath this article is models.json, our machine-readable price and free-tier index. Every figure on this page pulls from that source and carries its verification date. When a price event happens, both the source file and this article update in the same session.

The Live Pricing Table (Updated June 2026)

Claim: Every AI tool in this table has either changed its list price, degraded its free tier, or inflated its effective per-task cost in the first half of 2026. None of them are the same product they were in January.

Table sourced from claude.com/pricing2026-06-10, CloudZero ChatGPT pricing guide2026-06-10, and Fritz AI ChatGPT pricing breakdown2026-06-10. Figures reflect June 10, 2026 observations; prices can change without notice.

What Do Free Tiers Actually Give You?

Claim: The phrase "free tier" conceals wildly different access windows -- from a nearly unlimited search engine to a tightly throttled chat interface that degrades under load. Matching the tool to your actual use pattern matters more than whether it is technically "free."

Here is what each free tier actually delivers in June 2026, after we account for the changes that do not make it to the headline pricing page.

Claude Free gives you access to Claude Sonnet2026-06-10, a capable mid-tier model, with no published message cap -- but Anthropic's actual limit is dynamic. Under heavy server load the platform both slows responses and drops you to lighter models. You do not get Projects (which lets Claude maintain context across conversations), full Opus 4.8 access, or file uploads in all interfaces. No ads appear on Claude Free. This makes it the cleanest consumer free tier for chat right now.

ChatGPT Free changed materially in 2026. Sponsored ads began appearing on Free and Go tier sessions in the United States as of February 9, 20262026-06-10. Plus and Pro remain ad-free. The free model access (GPT-4o limited plus GPT-4o mini) did not change, but the flagship model on the platform shifted. GPT-5.5 launched April 23, 20262026-06-10 and is locked to the $100 and $200 Pro tiers, Business, and Enterprise. Free and Plus users remain on GPT-4.1-class models.

Gemini Free is the outlier on context window size -- Google offers up to 1 million tokens in context even on the free plan, which beats every other tool in the table. The catch is model access: Gemini 2.5 Pro (Google's strongest model) is paywalled behind the $19.99/month Advanced plan. Free users get Gemini 2.0 Flash, which is capable but not Google's flagship.

Perplexity Free is the best option if your primary use is search-augmented queries. The free tier offers unlimited standard searches and approximately five Pro searches per day, where Pro searches use a high-tier model (Claude Sonnet, GPT-4o, or Perplexity's own Sonar Pro) with live web retrieval. File uploads are not included on free. For pure information retrieval tasks, Perplexity Free delivers more per zero dollars than any of the chat-first alternatives.

The Free-Tier Nerf Log

Claim: Three dated pricing events in H1 2026 changed the value equation for consumer AI tools without a single published list-price increase. This is the log.

This log is the recurring-update hook that static listicles do not have. Every entry here is dated, sourced, and will be added to as new events occur. The source data file is nesyona-data/models.json, which gets updated on each price event.

Effective Price: The Number Pricing Pages Hide

Claim: The list price on a pricing page is not the cost of using an AI tool. Tokenizer inflation, model access restrictions, and free-tier degradation all change the real cost of a given task without appearing in the headline number.

We track effective price as a basket of 300 representative tasks per month (averaging 850 input tokens plus 600 output tokens per task at the model's list per-token rate, then multiplied by any tokenizer inflation factor). This is not a precise billing simulator -- it is an index for comparing direction of change across tools over time. The source methodology is in models.json.

The numbers here are modeled observations from our indexed data, not Anthropic or OpenAI billing records. The signal is the direction and magnitude of change, not the cents figure. Claude's effective cost increased more in absolute terms because tokenizer inflation is a real mechanism (more tokens billed per task) rather than a perceived shift. ChatGPT's effective Plus cost rise is smaller but is driven by a different dynamic: the free tier degraded enough (via ads, model restrictions) that quality-sensitive users upgrade, paying $20 for what previously came free.

Is Claude Free? The Real Answer

Claim: Claude is technically free but functionally limited in ways that the pricing page does not spell out. The question that matters is not "is it free" but "is the free tier adequate for your actual use pattern."

Yes, claude.com/pricing2026-06-10 lists a free tier. What it gives you is rate-limited access to Claude Sonnet -- a capable model, not a dummy placeholder. You can have real conversations, draft real documents, and get useful work done on it. What it withholds is reliability (rate limits spike under load), the top model (Opus 4.8 is Pro-only), and Projects (the memory system that makes Claude useful across days and sessions rather than session-by-session).

The more interesting question in 2026 is the "Claude free vs Pro" comparison after the tokenizer change. Upgrading to Pro at $20/month removes throttling and adds Opus access -- but you are paying for a product whose effective per-task cost rose 12 to 27% in April 2026 without a list-price change. Independent analysis2026-06-10 describes this as the "AI shrinkflation" pattern: same price, less output per dollar.

The answer to "is Claude free" is yes, with an asterisk. The asterisk is: no published message cap, dynamic throttling, no top model access, and the Pro upgrade now buys you more tokens of throttle-free access at an effective per-task cost that went up quietly in 2026. That is not a reason to avoid it -- Claude Sonnet on the free tier is genuinely good -- but it is a reason to understand what you are actually paying for if you upgrade.

Best Value Free Tier Pick

Claim: After the ChatGPT ad change, Claude Free is the cleaner free chat option for quality-conscious users. Perplexity Free wins for search-integrated queries. Neither requires a credit card.

We index this pick against the nerf log. The pick may change when a new price event lands.

ChatGPT Free is not our current free-tier pick for quality-sensitive users. The ads are not catastrophic, but they are a signal about OpenAI's monetization direction for free users. The model access on Plus ($20/month) is still strong -- GPT-5.5 restricted, but GPT-4o and GPT-4.1 class models remain -- and ChatGPT Plus remains competitive with Claude Pro for most general tasks. If you are evaluating the Plus tier against Claude Pro for paid use, the head-to-head pricing comparison runs through each use case in detail.

For teams evaluating AI tools at scale, the per-seat math changes substantially. If your team needs multiple seats, look at the business tier pricing -- Claude Teams, ChatGPT Business, and Gemini for Workspace all price by seat with different model-access floors. The colleagues at Lucreya's AI GTM stack cost breakdown runs the enterprise seat math for go-to-market teams specifically.

Bottom Line

The headline prices for consumer AI tools have been flat since 2023. The real prices have not. In the first half of 2026, Claude's effective per-task cost went up 12 to 27% via a tokenizer change that does not appear on any pricing page. ChatGPT's free tier acquired ads and its best model got locked behind a $100/month paywall. Gemini and Perplexity made fewer visible moves, but the platform landscape as a whole moved toward heavier monetization of free users and steeper gating of top model access.

This page -- backed by the models.json data layer -- updates on each price event. The free-tier alert in the email capture above exists specifically so you do not have to check manually. When the next pricing nerf drops, you will hear about it without revisiting this page.

The right call in June 2026: if you want free AI chat without ads, Claude Free remains the cleanest option. If you are evaluating the $20 paid tier between Claude and ChatGPT, the value proposition for Claude Pro narrowed slightly this year due to tokenizer inflation -- but it is still a capable product for users who need Opus 4.8 and Projects. For search-integrated tasks, Perplexity's free tier is structurally undervalued and gets little marketing attention.

Frequently Asked Questions