Price terms as published
Each line is the vendor’s own wording, split from the price capture checked on 2026-08-29. Nothing here is rounded, converted or re-worded — the full sentences stay in the review below. Lines that carry only a price continue the plan named on the line above, in the vendor’s own order.
| # | Plan / term | Checked |
|---|---|---|
| 1 | $20/mo | 2026-08-29 |
Who this comparison is for
You’re deciding what to do with roughly $20 a month across ChatGPT, Claude, or Gemini — and the vendor pages describe their paid tiers in three incompatible vocabularies: messages, rolling windows, weekly caps, “higher access.” Worse, the number that actually decides your experience (how often you hit a wall mid-task) isn’t in any of the headline tables. This piece puts all three on one set of axes: official prices as of 2026-08-29, how each vendor’s limit machinery behaves, and what’s quietly gated by region or tier.
One disclosure up front: the author of this site pays for ChatGPT Plus specifically for its coding agent, uses it daily for writing and reviewing generated code, and treats the chat surface as secondary. That lens shapes the verdict; the tables below do not.
The prices (official pages, checked 2026-08-29)
ChatGPT (OpenAI)
| Plan | Price | What it gets you |
|---|---|---|
| Free | $0 | Unlimited basic text chats, limited uploads/images/voice |
| Go | $8/mo | More messages/uploads/voice. Ads may appear depending on market and rollout status — not shown to all Go users everywhere |
| Plus | $20/mo | GPT-5.6 reasoning at Medium/High effort (Extra High and Sol Pro are Pro-tier), expanded Deep Research, Projects, custom GPTs, expanded Codex access |
| Pro | $100/mo (5×) or $200/mo (20×) | Maximum usage, Sol Pro reasoning tier |
A context-window caution we’d rather state than get wrong: OpenAI publishes different figures for Chat, Work, Codex, and API surfaces and third-party roundups mix them daily. Chat’s manual thinking mode is publicly documented around 256k; API-facing GPT-5.x models reach 1M on some listings. If long context is your requirement, verify which surface delivers it on your plan before trusting any table — including old versions of ours.
Claude (Anthropic)
| Plan | Price | What it gets you |
|---|---|---|
| Free | $0 | Chat on all platforms at low limits, 200k context |
| Pro | $20/mo (~$17/mo billed yearly) | Higher limits, Projects, Artifacts, Claude Code, web search |
| Max 5× | $100/mo | 5× Pro usage; top model access via usage credits |
| Max 20× | $200/mo | 20× Pro usage |
The number that matters is the denominator: Pro’s allowance refreshes in rolling 5-hour windows, and Opus-class usage burns a window several times faster than Sonnet — roughly 45 Sonnet messages per window against 15-25 Opus by community metering. On the API side Claude’s current flagships ship with 1M-token windows at standard pricing, which is the honest long-context answer none of the chat-plan tables show.
Gemini (Google)
| Plan | Price | What it gets you |
|---|---|---|
| Free | $0 | Everyday Gemini access with daily limits |
| AI Plus | $4.99/mo | 2× Free usage, 400 GB storage (repriced from $7.99 in June 2026) |
| AI Pro | $19.99/mo | ~4× usage vs Free, Gemini 3 Pro class models, Deep Research, Veo, 5 TB storage |
| AI Ultra | $99.99/mo (5×, 20 TB) or $199.99/mo (20×, 30 TB) | Highest limits, Gemini Spark (US-only), YouTube Premium |
Google restructured at I/O 2026 (cutting Ultra from $250) and refreshes limits every five hours with automatic downgrade to a lighter model at the cap rather than a hard stop. The sub-$5 entry tier is something neither rival matches; the tier naming (Plus/Pro/Ultra with storage bundles) is something neither rival needs to apologize for.
Where each vendor’s limit machinery actually hurts
This is the section the official tables don’t write, and it’s where buying decisions go wrong:
ChatGPT: the pain is layered meters. Plus’s Codex access runs on a five-hour window plus a weekly cap (OpenAI briefly dropped the 5-hour layer, then restored it on 2026-08-25 citing compute smoothing — which confirms the weekly number is the real budget). For steady daily coding you’ll live inside Plus fine; for burst weeks you’ll meet the wall and buy credits.
Claude: the pain is model-weighted windows. You don’t run out of “usage,” you run out of Opus usage first, mid-sentence, and the product quietly shifts you down. If you don’t care which model answers, Pro’s $20 is the calmest experience of the three.
Gemini: the pain is navigability. Per-app daily limits plus five-hour refresh plus a storage-and-benefits bundle means the plan you’re comparing against Plus isn’t really the same product category. If you live in Google Workspace the bundle is half your bill; if you don’t, you’re pricing 5 TB of Drive you won’t use.
Choosing by what you already pay for
- You already pay for Google One/Workspace: AI Pro stacks Veo and 5 TB onto subscriptions you hold; marginal value is cheapest here — and AI Plus at $4.99 is the price floor of the whole market.
- You code with a terminal daily: Claude Pro (Claude Code) and ChatGPT Plus (Codex) are the two serious options at $20; the weekly-cap math, not the model debate, should decide.
- You want one subscription for everything: ChatGPT Plus covers the most surfaces — provided you accept breadth over best-at-any-one.
- You write long documents and want the prose to stay yours: Claude’s editing behavior is the common answer, with the window caveat above.
Verdict
From someone who actually pays $20/month into this market — for ChatGPT Plus, and uses it through Codex to write and review code rather than to chat: Plus is the right pick for an agent-shaped workflow, and the same $20 is probably wasted if all you do is ask it questions, because the free tiers and Gemini’s $4.99 Plus cover question-asking honestly. Our hesitation in this comparison reflects the metering whiplash (a limit regime changed twice in one month) more than the product: strongest surface breadth, least stable terms.
Concretely, the day-to-day on this site: a separate agent writes the code, Codex reviews it before anything ships — author and auditor in different harnesses by design — and the weekly cap has never once run out before a normal day’s review rounds finished. That is the exact usage profile Plus is worth $20 for, and the exact profile where pure-chat buying is worth nothing.
Methodology & sources
Prices and tier terms checked by our editor 2026-08-29 against openai.com/chatgpt/pricing, anthropic.com/pricing, gemini.google/subscriptions, and one.google’s AI plan page; the Codex five-hour restoration from OpenAI’s 2026-08-25 announcement (9to5Mac coverage); Google’s I/O 2026 repricing from blog.google. The author’s ChatGPT/Codex usage is first-hand; Claude and Gemini sections document published terms, not hands-on benchmarks. Re-verify before acting — all three vendors changed pricing or limits within the last 90 days.