ZenkenAI
Published: Updated:

Claude Usage Limits Explained: Free, Pro, Max — All Plans (2026 Edition)

Claude
Table of contents 13 items
Zenken is featured as an OpenAI customer story

Short version: Claude is not truly unlimited. Every plan has some kind of cap, and the differences between plans are large. This article uses Anthropic’s official help center and changelogs to give you the picture as of June 2026.

The latest flagship — Claude Opus 4.8 — released May 28, 2026, is now available on Pro and above (the previous-generation Claude Opus 4.7 is still selectable). On May 6, 2026 Anthropic also doubled Claude Code’s 5-hour rate limits for Pro / Max / Team / seat-based Enterprise plans and removed peak-hour throttling for Pro and Max.

補足

Reset timing: 5-hour limits use a rolling window (5 hours from your first message). Daily limits reset at 00:00 UTC. Weekly limits reset 7 days from your first message at 00:00 UTC.

📊 Claude usage limits — quick reference (June 2026)

All plans — pricing and limits (June 2026)

PlanMonthlyMessagesClaude CodeCoworkContext
Free$0Dynamic❌❌200K
Pro$20~45 / 5 hrs40–80 hrs/wk✅1M
Max 5x$100~225 / 5 hrs140–280 hrs/wk✅1M
Max 20x$200~900 / 5 hrs240–480 hrs/wk✅1M
Team Standard$25/seatPer-seat✅✅1M
Team Premium$125/seat ($100/yr)Per-seat (6.25×)✅✅1M
EnterpriseContact salesCustom✅✅1M

Messages per 5-hour window by plan

Model access by plan (June 2026)

ModelFreeProMaxTeam / Enterprise
Claude Opus 4.8❌✅✅✅
Claude Opus 4.7❌✅✅✅
Claude Opus 4.6❌✅✅✅
Claude Sonnet 4.6✅✅✅✅
Claude Haiku 4.5✅✅✅✅
Claude Code❌✅✅✅
Cowork❌✅✅✅

Claude Opus 4.8 (May 2026)

Anthropic released Claude Opus 4.8 on May 28, 2026 — the latest and most capable generally available model (flagship), replacing the previous-generation Opus 4.7 at the top of the lineup. It is available on Pro and above, with model ID claude-opus-4-8.

  • Better coding accuracy: about one-quarter as likely to miss flaws in its own code compared with the previous generation (Opus 4.7)
  • Stronger agentic ability: scores 84% on the Online-Mind2Web browser-automation benchmark, ahead of both Opus 4.7 and GPT-5.5
  • Improved browser use, legal, finance, and search/citation: more accurate on practical, work-oriented tasks than the previous generation
  • Selectable effort: choose how deeply Claude reasons (effort level) in claude.ai and Cowork
  • Fast mode: up to 2.5× faster output. API pricing is $10 input / $50 output per 1M tokens — roughly one-third the price of the previous fast model
  • Dynamic Workflows (Claude Code, research preview): run hundreds of parallel sub-agents in a single session; available on Enterprise / Team / Max plans
補足

Standard API pricing for Opus 4.8 is $5 input / $25 output per 1M tokens — the same as the previous-generation Opus 4.7. Adaptive Thinking (inherited from Opus 4.7) is the only thinking mode, and effort can be set to low / medium / high / xhigh / max.

Availability: claude.ai (selectable on Pro and above), Claude API, Amazon Bedrock, Google Cloud Vertex AI, GitHub Copilot, and Claude Code.

Claude Opus 4.7 (April 2026, previous generation)

Anthropic released Claude Opus 4.7 on April 16, 2026 — the previous-generation model that was Anthropic’s most capable until May 2026. It is still selectable on Pro and above.

  • Major software-engineering gains: clear improvements over Opus 4.6 on hard coding tasks
  • Better instruction following: more reliable on detailed requirements and format specs
  • 3× vision performance: image long-edge up to ~2,576 px (~3.75 MP)
  • Stronger self-verification: Claude checks its own output before delivering on long tasks
  • New xhigh effort level: finer control over reasoning vs. speed
  • Task Budgets (public beta): cap usage on a per-task basis
  • /ultrareview command: dedicated code-review session in Claude Code

SWE-bench Verified score comparison (April 2026)

API pricing

$5 input / $25 output per 1M tokens — same as Opus 4.6. Model ID: claude-opus-4-7.

注意

The updated tokenizer can produce 1.0×–1.35× more tokens for the same input text. Measure your actual usage before scaling up to avoid unexpected costs.

Model release timeline

Claude model release history

ModelReleasedNotable for
Claude Opus 4.8May 28, 2026Latest flagship. Adaptive Thinking only, with Fast mode.
Claude Opus 4.7April 16, 2026Previous flagship. Adaptive Thinking only — no Extended Thinking.
Claude Sonnet 4.6February 17, 2026Big jumps on coding, knowledge work, agentic planning
Claude Opus 4.6February 5, 20261M context, Adaptive Thinking, Agent Teams
Claude Opus 4.5November 24, 2025effort parameter introduced
Claude Sonnet 4.5September 29, 2025Coding / agent specialization
Claude Opus 4.1August 5, 2025Stronger agent tasks
Claude Opus 4 / Sonnet 4May 22, 2025Generation-4 base models

Free plan

Free uses Claude Sonnet 4.6.

  • Short messages: ~40/day
  • Long messages or attachments: ~20–30/day
  • Reset: every 5 hours
メモ

As of May 2026, Free limits are dynamic — they vary based on server load and query complexity. During peak hours, Free users are deprioritized and may hit their cap sooner.

February 2026 Free expansion (effective Feb 11, 2026): File creation (Word, Excel, PowerPoint, PDF), third-party connectors (Slack, Canva, Notion, Figma), Skills, and Conversation compaction were all added to the Free plan.

Claude Pro ($20/month)

  • Messages: ~45 / 5 hours (~216/day)
  • Claude Code: 40–80 hours/week of Sonnet 4.6
  • Context window: 1M tokens
ポイント

How the rolling window works: The 5-hour window starts from your first message. Send your first message at 9 AM → fresh allowance at 2 PM. Message length, conversation history, and attachments all affect how many messages fit in one window.

Pro also includes: Opus 4.8 access (the previous-generation Opus 4.7 / 4.6 are still selectable), Claude Code, Cowork, MCP Apps, Projects, Google Workspace, Research, priority access, and the ability to buy extra usage at API rates.

Claude Max ($100–$200/month)

  • Max 5x ($100/mo): ~225 / 5 hours, 140–280 hrs/week Claude Code (Sonnet), auto-switches Opus → Sonnet at 20% usage
  • Max 20x ($200/mo): ~900 / 5 hours, 240–480 hrs/week Claude Code (Sonnet), auto-switches at 50% usage
メモ

Pro / Max usage is shared across Claude (chat), Claude Code, and Cowork. Whatever tool you use draws from the same counter.

Claude Code limits

Claude Code weekly allowance by plan

PlanSonnet 4.6Opus 4.8 / 4.7 / 4.6
Pro40–80 hrs/weekLimited
Max 5x140–280 hrs/week15–35 hrs/week
Max 20x240–480 hrs/week24–40 hrs/week

Claude Code — weekly hours (Sonnet 4.6, midpoint estimate)

注意

Claude (chat), Claude Code, and Cowork share a usage counter. Heavy use of Claude Code reduces what is left for chat and Cowork.

Context window

Context window by plan

Plan / modelContext windowApproximate equivalent
Free (Sonnet 4.6 / Haiku 4.5)200K tokens~150,000 words (~300 pages)
Pro / Max / Team / Enterprise (Opus 4.8/4.7/4.6/Sonnet)1M tokens~750,000 words (~1,500 pages)
補足

Context Compaction: Introduced with Opus 4.6 (beta). Long conversations are automatically summarized so you can keep going without losing key details when you approach the context limit.

Summary: which plan should you pick?

  1. 1

    Check your daily message frequency

    Fewer than 10 messages a day? Free may be sufficient. Use Claude heavily for work every day? Pro is the most cost-effective starting point.

  2. 2

    Decide which model you need

    Need Claude Opus 4.8 (the latest flagship)? You need Pro or above. If Sonnet 4.6 covers your work, the Free plan handles it.

  3. 3

    Check if you need Claude Code or Cowork

    Claude Code (AI coding agent) and Cowork (desktop AI agent) require at least the Pro plan. Free users cannot access either.

  4. 4

    Check if it's for a team

    Multiple users? Look at the Team plan. Need SSO, audit logs, and admin controls? Consider Enterprise.

ポイント

Most individual users land on Claude Pro ($20/month). ~45 messages per 5 hours and a 1M context window cover most daily business work. For pricing details see Claude Pricing Guide.

References