What Claude Code, Codex and Cursor really cost per project

Every paid Claude plan includes Claude Code. Web, desktop, mobile and Claude Code draw on the same usage.

• Free: no Claude Code.
• Pro: $20 a month, or $17 a month billed annually ($200 up front).
• Max: from $100 a month, with 5x or 20x Pro's usage per five-hour session.
• Team: standard seat $25 a month, or $20 a month billed annually; premium seat $125 or $100 a month, with 5x the usage of a standard seat. 2 to 150 seats.
• Enterprise: $20 per seat per month billed annually, plus usage at API rates.
• API (Claude Console): per million tokens, for example Sonnet 5.5 at $2 input and $10 output, Opus 5.5 at $4 input and $20 output, Haiku 5.5 at $0.10 input and $0.50 output (prompts up to 100,000 tokens). Cache reads cost a fraction of the input price, cache writes a little more.

Codex is included in ChatGPT Free, Go, Plus, Pro, Business, Edu and Enterprise.

• Go: $8 a month. Plus: $20 a month.
• Pro: three tiers at $100, $200 or $500 a month.
• Business: $25 per seat per month billed monthly, or $20 per seat per month billed annually for two seats or more.
• API key: Codex usage at API list prices, for example gpt-6.1-sol at $2 input, $0.10 cached input and $10 output per million tokens, or gpt-5.3-codex at $1.75, $0.175 and $14.

• Hobby: free, with limited agent requests.
• Pro: $20 a month. Pro Plus: $60 a month. Ultra: $200 a month.
• Teams: Standard $40 per seat per month, Premium $120 per seat per month with 5x the Standard agent limits.
• Included usage is counted at the API price of each model, so your model choice decides how fast it runs out. Auto bills at the list price of the model a request is routed to.
• Cursor's own Composer 2.5 costs $0.50 input and $2.50 output per million tokens. On Teams and Enterprise, third-party models add a Cursor Token Rate of $0.25 per million tokens.

A subscription caps usage instead of billing it. What happens at the limit differs by tool:

• Claude: limits reset on a rolling five-hour window, and paid plans add weekly limits. "You've hit your session limit" or "You've hit your weekly limit" names the reset time. Until then you wait, upgrade, or turn on usage credits, which continue at API rates. In Claude Code, /usage shows your plan usage bars.
• Codex: OpenAI estimates local messages per five-hour period for Plus and Business, depending on the model; Pro currently has no five-hour limit. Weekly limits may also apply, and cloud tasks use more of the allowance than local messages. Beyond the limit, Plus, Business and Pro continue with credits.
• Cursor: no time window. Once the included usage is spent, on-demand usage continues at the same API rates, billed monthly in arrears, or you move to a higher plan.

At API prices, an agent hour costs the sum of four token types times their list price. An example for an hour of active runtime with a lot of context, at Sonnet 5.5 prices:

• 0.3 million new input tokens × $2 = $0.60
• 15 million cache reads × $0.10 = $1.50 (the context is re-read with every request)
• 1 million cache writes (5 minutes) × $2.50 = $2.50
• 0.2 million output tokens × $10 = $2.00, thinking included
• Total: $6.60 per hour. The same hour on Opus 5.5 costs $13.20, as all its prices are twice as high.

The token amounts in the example are assumptions; long sessions, subagents and large codebases push them up. Anthropic puts the average Claude Code cost in enterprise deployments at around $13 per developer per active day and $150–250 per developer per month.

On a subscription, an extra session costs nothing until you hit the limit. The real cost per project is the plan price times the project's share of your usage. Divided by the project's agent runtime, that gives the cost of an agent hour, which you can compare with the API price.

/usage in Claude Code or a Codex usage dashboard shows what you consumed. Which client it was for, and what it cost you there, needs the numbers from the calculators for every project, every month.

Time Momentum does that automatically. It captures sessions from Claude Code, Codex and Cursor per project and values each one at the list price of its model that applied when the session started, converted to euros at the ECB rate of that day. On a Claude or ChatGPT subscription, it shares the monthly amount out by each project's share of the tokens instead. That makes it a Claude Code usage monitor per client, with your working time next to it.