As of November 2026, Claude Code is included in paid Claude plans: Pro costs $20 per month, Max costs $100 and Ultra costs $200, while API use is billed per token. Claude Code usage through Requesty averaged $131 per routing key in October 2026.
A plan can cost less than an API bill when comparable work fits its usage limits. If you enable paid usage credits, work beyond those limits is billed at API rates on top of the plan price.
What does Claude Code cost on each plan?
Anthropic includes Claude Code in its paid plans. Coding sessions and Claude chats share the plan's usage limits.
The table below shows November 2026 prices. Monthly figures are per month; annual figures are monthly equivalents paid on an annual bill.
| Plan | Monthly billing | Annual billing | Claude Code |
|---|---|---|---|
| Free | $0 | $0 | Not included |
| Pro | $20 | $17 per month | Included |
| Max | $100 | Monthly billing only | Included |
| Ultra | $200 | Monthly billing only | Included |
| Team Standard | $25 per seat | $20 per seat per month | Included |
| Team Premium | $125 per seat | $100 per seat per month | Included |
| Enterprise | Annual billing only | $20 per seat per month, plus API-rate usage | Included, with usage billed at API rates |
Source: Anthropic pricing.
Max gives 5 times Pro's usage every 5 hours and every week. Ultra gives 10 times Pro's weekly usage and has no 5-hour limit for verified accounts, according to Anthropic.
On October 30, 2026, Max 5x became Max and Max 20x became Ultra, with prices unchanged. Anthropic now describes Ultra as having 10 times Pro's weekly usage.
When you reach a plan limit, Anthropic gives you three options: wait for it to reset, move to a higher plan or enable paid usage credits at standard API rates.
Max also includes $100 per month in API credits to build on the Claude Platform, and Ultra includes $200, according to Anthropic.
What does Claude Code cost on the API?
What you pay on the API depends on the model and the number of tokens it processes. A token is a chunk of text the model reads as input or generates as output.
These are November 2026 rates per million tokens:
| Model | Input | Output |
|---|---|---|
| Haiku 5.5 (prompts up to 100,000 tokens) | $0.10 | $0.50 |
| Sonnet 5.5 | $2 | $10 |
| Opus 5.5 | $4 | $20 |
| Fable 5.2 | $8 | $40 |
| Fable 5.1 | $10 | $50 |
Source: Anthropic's pricing documentation.
Caching changes the bill. A cache read costs $0.10 per million input tokens for Sonnet 5.5 and $0.20 for Opus 5.5. A 5-minute cache write costs 1.25 times the base input rate, while a 1-hour write costs twice the base input rate, according to Anthropic.
Haiku has one threshold to watch: a prompt over 100,000 tokens is billed at 5 times the per-token rate, including cache reads and writes, according to Anthropic. Anthropic's Batch API discounts input and output tokens by 50%, while US-only inference has a 1.1 times multiplier.
For more on the model behind the Opus rates, see what changed with Opus 5.5.
What do Claude Code users spend per month?
Our coding agent cost data puts Claude Code at an average of $131 per user in October 2026. In this dataset, a user is a unique API routing key.
| Month in 2026 | Average monthly spend per routing key |
|---|---|
| April | $78 |
| May | $69 |
| June | $110 |
| July | $95 |
| August | $65 |
| September | $148 |
| October | $131 |
The average moved from $78 in April to $131 in October. That is an increase of $53, or about 68%.
Anthropic's Claude Code cost guidance uses a separate enterprise figure: about $13 per developer per active day, and $150 to $250 per developer per month. Anthropic also says 90% of users stay below $30 per active day. The two sets of figures cover different groups and use different definitions of a user.
How many tasks does it take to pass a plan price?
Our coding benchmark used Requesty's own agent on 129 private Go backend tasks. Sonnet 5.5 cost $0.52 per task and Haiku 5.5 cost $0.11.
To get an API-cost equivalent for each plan, we divided the monthly plan price by the benchmark cost per task. The calculation excludes paid usage credits, and task counts are rounded up to the first whole task that reaches or exceeds the plan price.
| Monthly plan price | Sonnet 5.5 at $0.52 per task | Haiku 5.5 at $0.11 per task |
|---|---|---|
| Pro: $20 | 39 tasks | 182 tasks |
| Max: $100 | 193 tasks | 910 tasks |
| Ultra: $200 | 385 tasks | 1,819 tasks |
For this comparison, assume that task costs stay at the benchmark average and that the same work fits inside the subscription's usage limits. Because the benchmark used Requesty's own agent on Go tasks, these results are an indication of API cost, not a count of Claude Code tasks included in a plan.
The cost per solved task is higher because some tasks are not solved. Sonnet solved 87.6% at $0.59 per solve, and Haiku solved 81.4% at $0.14 per solve. For work where a failed task needs another attempt, those differences can change the model choice.
Take a $131 monthly API bill: at the benchmark rate, it buys about 252 Sonnet tasks ($131 divided by $0.52). On a per-solve basis, it buys about 222 Sonnet solves.
Which plan should you pick?
If you have no Claude plan, compare matching billing periods: $20 for one month of Pro against one month of API charges, or $200 for a year of Pro billed up front ($17 per month) against a year of API charges. Paying per token is cheaper when the API total for that period is lower. If you already pay for a plan with room in its shared limits, coding inside that plan adds no extra cost, so compare extra API spending with the cost of an upgrade. For monthly API bills above $20, Pro is worth comparing if the work fits its usage limits; compare Max around $100 of API work per month and Ultra around $200. Use API billing when you want per-key spending caps or need to attribute usage to individual developers.
Take a hypothetical $131 monthly API bill for Claude-only work, with no Requesty fee included. If the same work fits Max's limits, its $100 base price is $31 less per month, or about 24% less. Ultra's $200 base price is $69 more.
The $131 figure matches our October average, but that observed average also includes non-Claude models. With Requesty's 5% fee, the assumed $131 provider bill becomes $137.55 ($131 times 1.05).
Apply Anthropic's $13 per active day: over 20 active days, that is $260 per developer, above Ultra's $200 monthly price. The subscription comparison still depends on whether that workload fits Ultra's limits or needs paid usage credits.
If you are also comparing coding tools, see Cursor vs Claude Code.
What does Claude Code cost for a team?
Team has Standard and Premium seats for groups of 2 to 150. A Standard seat has more usage than Pro, and a Premium seat has 5 times a Standard seat's usage, according to Anthropic.
Team also includes up to $500 per month in pooled API credits. Enterprise costs $20 per seat per month plus usage at API rates, is billed annually and adds SCIM, audit logs, role-based access and admin spend limits, according to Anthropic.
Take a team of 20 developers. Assume each developer uses Claude-only work that would cost $131 per month at API rates, and that the Requesty fee is 5% on top of that provider cost.
| Route | Monthly calculation | Cost |
|---|---|---|
| Team Premium, annual billing | 20 seats times $100 | $2,000 |
| Enterprise | 20 seats times $20, plus 20 times $131 in usage | $3,020 |
| API through Requesty | 20 times $131, plus 5% | $2,751 |
These figures buy different allowances. The Team Premium figure assumes work fits within the seats' plan limits; paid usage credits would add to the bill. The Enterprise figure includes API-rate usage, and the Requesty figure assumes the $131 is provider cost before the fee.
Enterprise has a 20-seat minimum. For volume and committed spend, Anthropic also has customized Enterprise and API pricing through sales.
For a setup across a team, see our guide to managing Claude Code costs per developer.
How do you control API spend?
Anthropic's cost guidance names an LLM gateway that tracks spend per key as a way to attribute Claude Code costs to users. Requesty can do this for API usage; Claude subscription usage cannot be routed through it.
- 1Create a key per developer
Give each developer a Requesty API key so you can attribute spending by key. Requesty also tracks costs by user, model, provider and time period.
- 2Connect Claude Code
Set
ANTHROPIC_BASE_URLandANTHROPIC_AUTH_TOKENin the shell or in Claude Code settings. Follow the values and setup in the Claude Code integration guide. - 3Set a monthly limit
Set a monthly spend limit per API key, user or group, and add budget alerts before the limit is reached.
- 4Review usage
Check spending by key and model, then adjust the cap for the developer's workload. Requesty's cost tracking docs cover the available breakdowns.
Requesty charges 5% on top of model costs.
How do you lower the API bill?
Before you increase the budget
0 of 5 done
The model with the lowest cost per task can leave more work for a developer to finish. Compare the solve rate as well as the price in our coding model benchmarks.
Start with one month's usage and a cap you can afford. If you use a subscription, check whether your coding sessions and Claude chats fit the shared limits. If you use the API, track cost per developer and revisit the model choice after you have task results.
Frequently asked questions
- How much does Claude Code cost per month?
- Claude Code is included in paid Claude plans. In November 2026, Pro costs $20 per month, Max costs $100 and Ultra costs $200. API use is billed per token.
- Is Claude Code free?
- Claude Code is not included in Claude's Free plan. You can use it with a paid Claude plan or pay for API usage.
- How much do Claude Code tokens cost on the API?
- November 2026 API rates for Claude Sonnet 5.5 are $2 per million input tokens and $10 per million output tokens. Opus 5.5 costs $4 per million input tokens and $20 per million output tokens.
- How much does the average Claude Code user spend?
- Claude Code usage through Requesty averaged $131 per routing key in October 2026. Anthropic reports $150 to $250 per developer per month across enterprise deployments.
- How much does Claude Code cost for Enterprise?
- Claude Enterprise costs $20 per seat per month, billed annually, plus usage at API rates. Self-serve Enterprise has a 20-seat minimum.
- Is Claude Max cheaper than using the API for Claude Code?
- Claude Max costs $100 per month. At the $0.52 per task measured for Sonnet 5.5 in Requesty's Go coding benchmark, comparable API costs pass $100 at 193 tasks. The plan comparison depends on whether the work fits Max's shared usage limits.
- What is the cheapest way to run Claude Code?
- If you have no Claude plan, compare matching billing periods: Pro costs $20 for one month, or $200 for a year billed up front ($17 per month). API usage costs less when your API total over that same month or year is lower. If you already have a paid plan with room in its shared limits, Claude Code adds no extra cost.
- OCT '26
Claude Code cost management for teams: Cursor and Codex in 2026
Claude Code cost management team guide: set Cursor and Codex limits, track subscriptions and API bills, and control coding agent spend across engineering.
- SEP '26
Cursor vs Claude Code (2026): Cost, Models and Team Controls
Cursor vs Claude Code: compare September 2026 plans, agent workflows, models, team controls and transparent monthly cost scenarios for developers.
- SEP '26
Claude Opus 5.5 API pricing, coding cost and speed
Claude Opus 5.5 API pricing: $4 per million input tokens and $20 per million output tokens, plus cost per solved coding task, cache prices and measured speed.
