Back to all plans
Claude
Anthropic official
Anthropic's flagship assistant with native Claude Code and Cowork—Opus 4.8 top-tier coding and 1M context
IDE ToolsClaude Subscription (incl. Claude Code)
SubscriptionToken API
Core models
Claude Opus 4.8Claude Sonnet 4.6Claude Haiku 4.5Claude Opus 4.7Claude Sonnet 4.5Claude Opus 4.6Claude Opus 4.5
Claude Opus 4.8
Current subscription flagship—adaptive thinking, 1M context, adjustable effort (incl. xhigh)—for complex reasoning, long-horizon agentic coding, and hard knowledge work on Pro/Max; heavy Opus use fits Max tiers.
Claude Sonnet 4.6
Current Sonnet workhorse—best speed/intelligence balance, 1M context, extended/adaptive thinking—what most developers default to; more quota-friendly than Opus.
Claude Haiku 4.5
Current fastest Haiku tier, 200K context—on Free with lower caps; Pro/Max raise limits for quick Q&A and light first-pass tasks.
Claude Opus 4.7
Previous Opus generation still under More models—close to 4.8 but not the default pick; heavy Opus work is better on Opus 4.8 or Max tiers.
Claude Sonnet 4.5
Sonnet 4.5 remains under More models—prefer Sonnet 4.6 for daily work; 4.5 suits legacy behavior or A/B testing.
See the official site for more models
Additional core model names still appear above, with full details on the latest official page.
Plan details
Pro
Everyday productivityRecommendedPrice
$20/mo
$20 monthly · $200/year (~$17/mo)
Monthly
$20/mo
Auto annual
$17/mo
$200 billed upfront annually
Pro
Everyday productivityRecommendedPrice
$20/mo
$20 monthly · $200/year (~$17/mo)
Monthly
$20/mo
Auto annual
$17/mo
$200 billed upfront annually
Usage
Higher message, file, Research, and model limits than Free—suited as a stable daily driver for writing, coding, document work, and light-to-medium agent tasks. Free is Haiku + Sonnet only; Pro unlocks the full Opus family.
Models
Includes Claude Code CLI, Claude Cowork, and Claude Design plus Projects, Research, Memory, Connectors, and more—Pro unlocks Opus 4.8, Sonnet 4.6, and Haiku 4.5 as primary models, switchable in claude.ai or Claude Code
/model.Highlights
Also includes Claude for Microsoft 365/Outlook integrations and Research mode—most individual developers treat Pro as the default chat + terminal coding + project organization bundle.
Usage is governed by a 5-hour rolling session window plus weekly caps (shared across claude.ai, Claude Code, and Cowork)—Haiku is lightest, Sonnet the default workhorse, Opus for hard tasks but burns fastest; long multi-turn context accelerates limit hits.
If you do not yet need Max’s 5x/20x usage, peak priority, or higher output ceilings, Pro is usually the best-value paid tier. Track Current session and Weekly limits progress at claude.ai/settings/usage.
Usage is governed by a 5-hour rolling session window plus weekly caps (shared across claude.ai, Claude Code, and Cowork)—Haiku is lightest, Sonnet the default workhorse, Opus for hard tasks but burns fastest; long multi-turn context accelerates limit hits.
If you do not yet need Max’s 5x/20x usage, peak priority, or higher output ceilings, Pro is usually the best-value paid tier. Track Current session and Weekly limits progress at claude.ai/settings/usage.
Best for
Frequent individual users, everyday primary coders, and indie developers needing Claude Code
Usage
Everything in Pro plus roughly 5x usage and higher output limits—suited to long daily Claude Code sessions, parallel projects, and sustained Opus 4.8 deep work.
Models
Includes priority access at high traffic times and early access to advanced features—Settings > Usage shows remaining Opus-only vs all-models weekly caps and reset times.
Highlights
Better for continuous large-repo analysis, heavy uploads, long Research chains, and individuals treating Claude as a semi-production environment—not occasional premium feature trials.
Deal
Team premium seats offer a similar ~5x standard-seat quota per seat in org plans; Max 5x is the personal-side equivalent.
Best for
Heavy individual developers, long-session researchers, and daily high-frequency Claude Code users
Usage
Everything in Pro plus roughly 20x usage—for parallel projects, all-day intensive coding/research, and near-continuous Claude Code workflows—Anthropic’s top individual tier for near-unlimited personal use.
Models
Also includes higher output limits, peak priority, and early feature access—suited to using Claude as a sustained production environment rather than an occasional assistant.
Highlights
The top choice for extreme individual users, parallel multi-repo operators, and long-horizon Opus 4.8 agent work—not a substitute for Team admin seats.
If Max 5x still hits limits or you need sustained Opus access during peak times, 20x is Anthropic’s tier for near-unlimited individual use.
If Max 5x still hits limits or you need sustained Opus access during peak times, 20x is Anthropic’s tier for near-unlimited individual use.
Best for
Extreme individual users running all-day writing, coding, and research with Claude as a primary production workspace
Quota
Product type
IDE Tools
Official product line
Claude Subscription (incl. Claude Code)
5-hour limit
5h rolling session window · shared by claude.ai, Claude Code, Cowork · Opus burns fastest
Weekly limit
Opus-only weekly cap + all-models weekly cap (independent resets) · Max 5x/20x ≈ 5×/20× Pro
Monthly limit
No fixed monthly msg count · session/week metering · Usage credits for overage
Notes
- US individual list prices: Pro $20/mo or $200/year (~$17/mo); Max from $100/mo (5x at $100, 20x at $200). Tax excluded—confirm at checkout.
- Team standard seats: $20/seat/mo annual or $25 monthly; premium seats (~5x standard usage): $100/seat/mo annual or $125 monthly. 5-seat minimum; mix standard and premium.
- All tiers opt out of model training on your content by default; Enterprise adds SCIM, audit logs, HIPAA-ready options, and custom data retention.
- claude.ai and Claude Code default to Opus 4.8, Sonnet 4.6, and Haiku 4.5; older versions remain under More models—deprecated models are excluded.
- Official limits use a dual system—5-hour rolling session window plus weekly caps—shared across claude.ai, Claude Code, and Cowork. Opus burns fastest, Sonnet moderate, Haiku lightest. Check Current session and Weekly limits (Opus-only vs all-models reset separately) at claude.ai/settings/usage.
- Public help docs do not publish fixed message counts (they vary with model, context length, and load). After hitting caps, wait for window resets or enable Usage credits on Pro/Max/Team for overage (see Anthropic support articles on usage credits).
Supported coding tools
Claude CodeClaude CoworkClaude DesignCursorDesktop AppSlackMicrosoft 365
Pricing and model data sourced from official vendor websites
FAQ
General8 条