The cheapest paid AI subscriptions in my data are Google AI Plus at $4.99 per month, Meta AI Core at $7.99 and ChatGPT Go at $8. The mainstream tier is about $20 per month, where ChatGPT Plus, Claude Pro, Google AI Pro, Cursor Pro and Microsoft 365 Premium sit. A subscription price does not tell you how much AI you get, and a flat fee does not cover API calls. This guide compares plans from official pages as of 2026-10-11 (USD) and shows where per-task API costs come in.
Cheapest paid tier by product
Most products below have a free tier, so the cheapest paid plan is the first step up. Perplexity is missing because my data lists no monthly price for Perplexity Pro or Max. Check its official pricing page.
| Product | Cheapest paid plan | Price per month | Next plan up |
|---|---|---|---|
| ChatGPT | Go | $8 | Plus, $20 |
| Claude | Pro | $20 | Max 5x, $100 |
| Google AI | Plus | $4.99 | Pro, $19.99 |
| Grok | SuperGrok | $30 | SuperGrok Plus, $100 |
| Meta AI (Meta One) | Core | $7.99 | Premium, $19.99 |
| Le Chat (Mistral Vibe) | Pro | $14.99 | Team, $24.99 |
| Microsoft 365 | Premium | $19.99 | None listed in my data |
| GitHub Copilot | Pro | $10 | Pro+, $39 |
| Cursor | Pro | $20 | Pro+, $60 |
Three things stand out:
- Google AI Plus is the cheapest paid plan. At $4.99 per month it gives 2x higher usage limits than Free, 200 Google Flow Credits and 400 GB of cloud storage.
- ChatGPT Go and Meta AI Core are the cheapest entry points among the big names. Go includes GPT-6 Luna at Standard speed in the desktop app, subject to rollout. Core adds more image and video generation with Meta AI.
- $20 is the common mainstream price. ChatGPT Plus, Claude Pro and Cursor Pro cost $20. Google AI Pro and Microsoft 365 Premium cost $19.99.
Top tiers
Top tiers cost many times the entry plans. Here is what the highest prices buy:
- ChatGPT Pro 500 at $500 per month has the highest included usage of the three Pro plans. It is the only one with Ultrafast mode (GPT-6 Astra Ultrafast, GPT-6.1 Sol Ultrafast). Pro 100 at $100 and Pro 200 at $200 leave Ultrafast out, and Pro 200 includes more usage than Pro 100.
- Claude Max 20x at $200 per month gives 20x more usage than Pro and $200 per month in API credits for the Claude Platform. Fable is limited to 50% of weekly limits on both Max tiers.
- Google AI Ultra (20x) at $199.99 per month gives 20x higher usage limits than AI Pro, plus first access to Deep Think and Gemini Spark in select countries.
- Grok SuperGrok Plus at $100 per month adds 1080p video, significantly higher usage across Chat, Imagine, Voice and Build, and priority access at peak times.
- Cursor Ultra at $200 per month and GitHub Copilot Max at $100 per month are the top coding tiers. Copilot Max includes $200 per month in total GitHub AI Credits.
A top tier makes sense when you hit the limits of the plan below it. It rarely makes sense as a first purchase.
Subscriptions versus API costs
A subscription covers the app. It does not cover the API. OpenAI states that with ChatGPT Plus, API usage is separate and billed separately. If you build on a model, you pay per token, and prices differ a lot between models.
For a simple chat answer, the median of all 326 models is 0.24¢ per task. The cheapest results are GPT-6 Luna at 0.06¢ and Claude Haiku 5.5 at 0.07¢. The most expensive is Claude Fable 5.1 at 6.3¢.
| # | Model | Cost per task | Text score | Tokens in / out | Basis | |
|---|---|---|---|---|---|---|
| 1 | GLM 5.3 FlashZ.ai | 0.04¢$0.36 / 1,000 | 1474 | 48 / 1,782 | measured best value | |
| 2 | GPT-6 LunaOpenAI | 0.06¢$0.59 / 1,000 | 1443 | 45 / 1,179 | measured | |
| 3 | Claude Haiku 5.5Anthropic | 0.07¢$0.66 / 1,000 | - | 61 / 1,299 | measured | |
| 4 | Qwen3.8 FlashAlibaba Qwen | 0.08¢$0.76 / 1,000 | - | 88 / 1,598 | measured | |
| 5 | Mistral Small 4Mistral | 0.08¢$0.83 / 1,000 | - | 40 / 1,377 | estimated | |
| 6 | MiMo-V2.6-ProXiaomi | 0.09¢$0.90 / 1,000 | 1479 | 206 / 933 | estimated best value | |
| 7 | DeepSeek V4.1 FlashDeepSeek | 0.09¢$0.91 / 1,000 | 1475 | 65 / 6,320 | measured | |
| 8 | Llama 4 MaverickMeta | 0.11¢$1.06 / 1,000 | 1327 | 87 / 1,592 | estimated |
List prices explain part of that. Claude Fable 5.1 lists at $10.00 input and $50.00 output per 1M tokens. Claude Haiku 5.5 lists at $0.10 and $0.50. Per chat answer, Fable costs about 90 times as much as Haiku (6.3¢ against 0.07¢).
Reasoning tokens are billed as output tokens, so totals can exceed what the visible answer suggests. DeepSeek V4.1 Flash used 5,461 reasoning tokens against 859 output tokens in my measured run, and its 0.76¢ per task is higher than its low list price suggests.
Two caveats from my methodology:
- Tokenizers differ. For the same English document, Anthropic's tokenizer produces 1.39 times the tokens of OpenAI's, so price per million tokens does not compare cleanly across labs.
- Some numbers are estimates. Twelve models were measured with one run per task, default settings, via OpenRouter, with output capped at 16,000 tokens. Other results are estimated.
Which plan for which user
Pick based on the job, not the brand.
- Casual users: Start free. Google AI Plus at $4.99 or Meta AI Core at $7.99 are the cheapest upgrades.
- Students and general writers: ChatGPT Go at $8, or Plus at $20, which adds Codex on the web, CLI, IDE extension and iOS. Usage can be extended with ChatGPT credits.
- Claude users: Claude Pro at $20 includes Claude Code. Fable is available only via usage credits on Pro. Max 5x at $100 gives 5x more usage than Pro and $100 per month in API credits.
- Developers who code daily: GitHub Copilot Pro at $10 includes $15 per month in GitHub AI Credits ($10 base plus $5 flex). Cursor Pro at $20 adds extended Agent limits, frontier models and cloud agents. Step up to Copilot Pro+ at $39 or Cursor Pro+ at $60 only if you hit limits.
- Heavy research or media work: ChatGPT Pro 100 at $100 includes Pro models, deep research, image creation, memory and file uploads. Google AI Pro at $19.99 gives 4x higher usage limits than Free and 5 TB of storage.
- Teams: Claude Team Standard seats at $25 cover 2 to 150 people and include up to $500 per month in pooled API credits. ChatGPT Business Standard seats at $25 require at least two paid seats and include SAML SSO and MFA. Le Chat Team at $24.99 adds domain name verification and data export.
- Builders of AI features: A subscription will not cover your product. Estimate API spend with the calculator and the chatbot answer task page.
For a quick look at model prices, see GPT-6 Luna and Claude Haiku 5.5.
FAQ
Is the cheapest paid AI plan worth it?
If free limits keep running out, yes. Google AI Plus at $4.99 and Meta AI Core at $7.99 are the lowest-priced paid options in my data. Check plan details on each provider's page first.
Does a $20 subscription include API access?
Not generally. OpenAI states ChatGPT Plus API usage is separate and billed separately. Claude Max plans include monthly API credits for the Claude Platform ($100 for Max 5x, $200 for Max 20x), but Claude Pro does not list them.
Why do per-task costs vary so much between models?
Price per token, output length and reasoning tokens all matter. A model with a low list price can cost more per task if it reasons at length. My chat answer data ranges from 0.06¢ to 6.3¢ per task.