TL;DR (June 2026): Claude's consumer plans are Free ($0), Pro $20/mo (or $17/mo billed annually, $200 upfront), Max at $100/mo (5x Pro) and $200/mo (20x Pro), Team at $25/seat/mo (Standard) and $125/seat/mo (Premium), and Enterprise at custom pricing. These are flat subscriptions for Claude.ai and Claude Code. They are not the same thing as the pay-per-token Claude API, which bills by usage. If you are a developer comparing a flat sub against per-token cost, the break-even section below is the part you want.
Claude Pro and the full plan lineup in 2026
"Claude Pro" usually means the $20/month individual subscription, but it sits inside a wider consumer lineup. Here is every plan, verified against Anthropic's pricing page in June 2026. The default model across paid plans is Claude Sonnet 4.6.
| Plan | Price (monthly) | Annual | What you get |
|---|---|---|---|
| Free | $0 | $0 | Claude Sonnet 4.6 (default) and limited Haiku 4.5, small daily usage |
| Pro | $20/mo | $17/mo ($200 upfront) | Much higher usage, Projects, web search, and Claude Code on a usage allowance |
| Max 5x | $100/mo | - | 5x Pro usage, priority access to new models and features |
| Max 20x | $200/mo | - | 20x Pro usage, highest individual Claude Code allowance |
| Team Standard | $25/seat/mo | $20/seat/mo | 5 to 150 seats, SAML SSO, central billing, admin controls, no training on your data by default |
| Team Premium | $125/seat/mo | $100/seat/mo | 5x Team Standard usage per seat |
| Enterprise | Custom | Custom | Volume seats, advanced security and compliance, custom limits |
The annual Pro discount is the only sub-$20 path to a paid plan: $17/month works out to $200 billed once a year, versus $240 if you pay monthly.
Claude Pro vs Claude Code vs the API: do not confuse the three
This is where most "claude pro price" searches go wrong, because three different products share the name.
- Claude Pro / Max are flat consumer subscriptions for using Claude.ai (web, desktop, mobile) and running Claude Code against a usage allowance. Predictable monthly bill.
- Claude Code is Anthropic's coding agent. It can run on a Pro/Max subscription allowance or on pay-per-token API billing. We break down the per-token side in what Claude Code actually costs in 2026.
- The Claude API is pay-per-token, billed by usage, for developers building applications. A Pro subscription does not include API access, and API spend has no flat cap unless you set one.
When the flat $20 (or $100) beats pay-per-token
For interactive chat and everyday coding help, the subscription almost always wins: a Pro user sending a few hundred messages a day would pay far more than $20 buying the same tokens on the API. The flat rate is effectively a heavy-usage discount with a predictable bill.
It flips for programmatic and agentic workloads. If you are running repository-wide Claude Code loops or driving Claude through the API in automation, usage can blow past any subscription allowance, and you are back on per-token economics where a runaway agent can produce a four-figure bill in a day. The practical rule in 2026:
- Light, interactive use: Pro at $20/mo. Don't touch the API.
- Heavy Claude Code use: Max 5x ($100) or 20x ($200) usually beats per-token, with a fixed bill.
- Building an app or automation: you need the API, and you need a spend cap and usage tracking, because the bill is variable by design.
The visibility gap: the moment you move from a flat Claude Pro sub to the usage-based API, your cost stops being a known $20 line and becomes a meter that can spike. UsageBox tracks usage-based AI spend per model and per project so you can see, in real time, whether a flat Pro or Max subscription would actually be cheaper than what your API usage is costing - and catch overruns before the invoice does.
What changed recently
Two shifts matter for anyone pricing Claude in 2026. First, Sonnet 4.6 became the default paid model in February 2026, so "Pro" buys more capable default access than it did a year ago. Second, Max plans absorbed most serious Claude Code usage, which is why heavy coding users increasingly pick Max over stacking API spend. Prices themselves (Pro $20, Max $100/$200) have held steady.