ChatGPT Plus and Claude Pro both cost $20 per month, so the choice depends on what each includes and how you work. Claude Pro includes Claude Code. ChatGPT Plus includes Codex on the web, in the CLI, in the IDE extension, and on iOS. The official pages do not publish message counts for either plan, so I compare what each plan includes and relate it to API-equivalent costs from my own measurements.
What each plan includes
Both plans sit at the same price point, with free tiers below them. The differences are in the extras and the upgrade paths.
| Plan | Price per month | What the official pages say |
|---|---|---|
| ChatGPT Free | $0 | Codex with GPT-6 Luna at Standard speed in the desktop app, subject to rollout |
| ChatGPT Plus | $20 | Codex on web, CLI, IDE extension and iOS; usage can be extended with ChatGPT credits; API billed separately |
| ChatGPT Pro 100 | $100 | Pro models, Codex, deep research, image creation, memory and file uploads; Ultrafast not included |
| Claude Free | $0 | Chat on web, desktop and mobile; no Opus or Fable access; no Claude Code |
| Claude Pro | $20 | More usage than Free; Claude Code included; Fable available only via usage credits |
| Claude Max 5x | $100 | 5x more usage than Pro; $100/month in API credits; Fable limited to 50% of weekly limits |
The key detail for Claude Pro is that Fable is not included in the subscription. You reach it only through usage credits. ChatGPT Plus usage can also be extended with credits, but the official page does not say a specific model is gated behind them.
The plan details give no message caps or token budgets for Plus or Pro. "More usage than Free" is the only usage claim for Claude Pro, so I cannot say which plan gives more messages per day. If that number decides it for you, check the live limits before you pay.
What the same work costs through the API
A subscription is good value only if the work you would otherwise pay for per token costs more than the flat fee. To estimate that, I use the API-equivalent cost of the same tasks. Some figures are measured and some are estimated, as labeled. See the methodology for details.
Writing: a 1,500-word blog post
The median across all 336 models is 0.87¢ per task. Selected models:
| Model | Cost per task | Per 1,000 tasks | Basis |
|---|---|---|---|
| GPT-6 Luna | 0.12¢ | $1.19 | measured |
| Claude Haiku 5.5 | 0.19¢ | $1.89 | measured |
| GPT-6.1 Sol | 2.3¢ | $23.00 | measured |
| Claude Sonnet 5.5 | 3.7¢ | $36.58 | measured |
| Claude Opus 5.5 | 7.3¢ | $73.16 | estimated |
| # | Model | Cost per task | Text score | Tokens in / out | Basis | |
|---|---|---|---|---|---|---|
| 1 |  DeepSeek V4.1 FlashDeepSeek | 0.04¢$0.37 / 1,000 | 1475 | 81 / 3,039 | measured best value | |
| 2 |  GPT-6 LunaOpenAI | 0.12¢$1.19 / 1,000 | 1443 | 59 / 2,360 | measured | |
| 3 |  GLM 5.3 FlashZ.ai | 0.14¢$1.39 / 1,000 | 1474 | 62 / 6,923 | measured | |
| 4 |  Claude Haiku 5.5Anthropic | 0.19¢$1.89 / 1,000 | - | 90 / 3,760 | measured | |
| 5 |  MiMo-V2.6-ProXiaomi | 0.20¢$2.04 / 1,000 | 1479 | 220 / 2,232 | estimated best value | |
| 6 |  MiniMax M3MiniMax | 0.27¢$2.72 / 1,000 | 1441 | 219 / 2,216 | measured | |
| 7 | GPT-5.4 NanoOpenAI | 0.30¢$2.96 / 1,000 | 1401 | 59 / 2,360 | estimated | |
| 8 |  Command A+Cohere | 0.34¢$3.40 / 1,000 | - | 219 / 2,221 | estimated | |
| 9 |  Mistral Small 4Mistral | 0.47¢$4.67 / 1,000 | - | 55 / 7,769 | estimated | |
| 10 |  Gemini 3.1 Flash LiteGoogle | 0.54¢$5.39 / 1,000 | - | 52 / 3,585 | estimated | |
| 11 | DeepSeek V4 Pro 0813DeepSeek | 0.61¢$6.07 / 1,000 | - | 81 / 3,039 | estimated | |
| 12 |  Qwen3.8 FlashAlibaba Qwen | ≥0.75¢$7.54 / 1,000 | - | 104 / 16,000 | measured capped |
A blog post costs a few cents through the API, even on the flagship models. Someone drafting a few posts a week would pay far less than $20 at API rates. A subscription pays off for writing only if you run many long sessions every day, or if you value the chat interface and extras more than the raw tokens.
Summarizing: a 20-page PDF
The median across all 336 models is 0.83¢ per task.
| Model | Cost per task | Per 1,000 tasks | Basis |
|---|---|---|---|
| GPT-6 Luna | 0.16¢ | $1.57 | measured |
| Claude Haiku 5.5 | 0.22¢ | $2.20 | measured |
| GPT-6.1 Sol | 3.1¢ | $31.05 | measured |
| Claude Sonnet 5.5 | 4.5¢ | $44.58 | measured |
| Claude Opus 5.5 | 8.9¢ | $89.16 | estimated |
Claude models bill more input tokens for the same PDF. The Claude Haiku 5.5 run used 18,004 input tokens, against 12,987 for GPT-6 Luna. Anthropic's tokenizer ratio is 1.39 against 1.00 for OpenAI on the same English text, which is part of why Claude tasks cost more here.
| # | Model | Cost per task | Text score | Tokens in / out | Basis | |
|---|---|---|---|---|---|---|
| 1 | DeepSeek V4.1 FlashDeepSeek | 0.09¢$0.85 / 1,000 | 1475 | 12,970 / 1,992 | measured best value | |
| 2 | GLM 5.3 FlashZ.ai | 0.10¢$0.99 / 1,000 | 1474 | 12,995 / 1,814 | measured | |
| 3 | GPT-6 LunaOpenAI | 0.16¢$1.57 / 1,000 | 1443 | 12,987 / 533 | measured | |
| 4 | Claude Haiku 5.5Anthropic | 0.22¢$2.20 / 1,000 | - | 18,004 / 790 | measured | |
| 5 | Qwen3.8 FlashAlibaba Qwen | 0.32¢$3.21 / 1,000 | - | 13,119 / 2,642 | measured | |
| 6 | GPT-5.4 NanoOpenAI | 0.33¢$3.26 / 1,000 | 1401 | 12,987 / 533 | estimated | |
| 7 | Mistral Small 4Mistral | 0.33¢$3.31 / 1,000 | - | 13,373 / 2,170 | estimated | |
| 8 | GLM 5.3Z.ai | 0.36¢$3.57 / 1,000 | 1478 | 12,995 / 1,814 | estimated best value | |
| 9 |  Llama 4 MaverickMeta | 0.42¢$4.16 / 1,000 | 1327 | 13,015 / 2,638 | estimated | |
| 10 | MiniMax M3MiniMax | 0.47¢$4.72 / 1,000 | 1441 | 12,901 / 708 | measured | |
| 11 | Command A+Cohere | 0.49¢$4.95 / 1,000 | - | 12,939 / 709 | estimated | |
| 12 | Gemini 3.1 Flash LiteGoogle | 0.51¢$5.06 / 1,000 | - | 12,741 / 1,248 | estimated |
One summary is cheap. If you process many PDFs a day with a flagship model, a flat fee starts to look better.
Coding
I have no cost figures for the code review task in this round, so I will not guess an API equivalent for coding. What I can compare is what each plan gives you:
- Claude Pro includes Claude Code.
- ChatGPT Plus includes Codex on the web, in the CLI, in the IDE extension, and on iOS.
- ChatGPT Free includes Codex with GPT-6 Luna at Standard speed in the desktop app, subject to rollout.
If you work in an IDE, Plus offers Codex there. If you work from a terminal, Claude Code on Pro and Codex in the CLI on Plus both cover it. For agent-style work, see the 50-step agent task.
Which subscription gives more value
- Choose ChatGPT Plus if you want one $20 plan that covers chat and coding across the web, terminal, IDE and phone. Credits can extend usage if you hit a limit. For everyday writing and summarizing, the measured GPT-6 Luna costs are low.
- Choose Claude Pro if terminal coding is your main use, since Claude Code is included. It suits you if you prefer Claude models and do not need Fable regularly. Fable on Pro means paying for usage credits.
- Look at the $100 tier if you hit limits daily on the biggest models. Claude Max 5x gives 5x more usage than Pro and includes $100/month in API credits. ChatGPT Pro 100 includes Pro models and Codex. Both cost $100, so pick the ecosystem you already use.
- Stay on a free tier if you only need occasional help. At a few cents per task, light use costs very little even through the API.
For most everyday professionals, pick the $20 plan that matches the tool you already use. Switching costs one month, so testing both is cheap. To price your own workload, use the calculator, and compare more subscriptions on the plans page.
FAQ
Is ChatGPT Plus or Claude Pro cheaper?
Both cost $20 per month. The difference is in what they include, such as Claude Code with Pro and Codex in several places with Plus.
Does Claude Pro include Fable?
Not as part of the subscription. The official page says Fable is available only via usage credits on Pro. Claude Max 5x and 20x limit Fable to 50% of weekly limits.
How much does a blog post cost through the API?
In my measurements, Claude Sonnet 5.5 cost 3.7¢ per blog post and GPT-6 Luna cost 0.12¢. These are per-task API costs, not subscription costs.
Sources
- OpenRouter model list and provider endpoints (list prices, read daily)
- ChatGPT plans (official page)
- ChatGPT plans (official page)
- ChatGPT plans (official page)
- Claude plans (official page)
- Claude plans (official page)
- Google AI plans (official page)
- Grok plans (official page)
- Perplexity plans (official page)
- Microsoft 365 plans (official page)
- Le Chat (Mistral Vibe) plans (official page)
- DeepSeek plans (official page)
- Meta AI (Meta One) plans (official page)
- Cursor plans (official page)