CodingPlans
← All plans
[ Anthropic ]

Claude

Flat-rate Claude Code access with opaque, relative limits.

Pro and Max give access to Claude Code and the full model range through 5-hour and weekly limits. Anthropic publishes limits only as multiples of Pro, so the dollar value of each plan has to be inferred from usage data.

Official pricing page ↗
[ Plans ]

Plan ladder

Values follow your controls on the Overview page (full-plan case).

Claude Pro

$20/mo

Est. · low
API value
$250
Value ×
13x
Output tok/mo
4.6M

5-hour session limit + weekly limit; Opus and Sonnet in Claude Code. Fable only via usage credits.

$17/mo if billed annually. Anthropic does not publish token or dollar allotments, only multiples (Max = 5x / 20x Pro).

Best model: Claude Opus 5.5 (AA 58) · source

Claude Max 5x

$100/mo

Est. · low
API value
$1,500
Value ×
15x
Output tok/mo
27.8M

5x Pro per-session allowance + weekly cap. Fable at 50% of weekly limits. Priority access at peak.

Also includes monthly API credits (amount not verified, not counted here).

Best model: Claude Opus 5.5 (AA 58) · source

Claude Max 20x

$200/mo

Est. · low
API value
$5,000
Value ×
25x
Output tok/mo
92.6M

20x Pro per-session allowance + weekly cap. Fable at 50% of weekly limits. Priority access at peak.

Widest uncertainty of any plan here: heavy agentic users report far more API-equivalent usage than light users. Use the override box with your own numbers.

Best model: Claude Opus 5.5 (AA 58) · source

Strengths

  • Top of the Artificial Analysis Intelligence Index (Opus 5.5 at 58, Sonnet 5.5 at 56)
  • Flat-rate Max tiers can carry a large subsidy for heavy agentic use (our values are estimates, so check against your own usage data)
  • Usage credits let you overflow at standard API rates instead of being cut off

Watch out for

  • Dollar allotments are unpublished; our values are estimates with wide ranges
  • Weekly caps and capacity-based limits can change without notice
  • Fable access on Pro/Max is limited (usage credits, or 50% of weekly limits on Max)
[ History ]

Value over time

Dated reference points, a snapshot ledger you extend with npm run snapshot, and the plan changes that moved the numbers.

Claude Pro: snapshot ledger (full-plan case)
0.0x7.2x14.4xOct 2026Nov 2026API value ÷ price12.5x

One snapshot so far. Each npm run snapshot adds a point.

Claude Max 5x: snapshot ledger (full-plan case)
0.0x8.6x17.3xOct 2026Nov 2026API value ÷ price15.0x

One snapshot so far. Each npm run snapshot adds a point.

Claude Max 20x: snapshot ledger (full-plan case)
0.0x14.4x28.7xOct 2026Nov 2026API value ÷ price25.0x

One snapshot so far. Each npm run snapshot adds a point.

What a fixed agentic workload costs at list price, by model generation

Opus: $ per 1M output tokens (all-in)
Opus 4 / 4.1
mid-2025
$233
Opus 4.5 to 5
Nov 2025 onward
$77.50
Opus 5.5
2026
$54
Sonnet: $ per 1M output tokens (all-in)
Sonnet 4 to 4.6
2025 to early 2026
$46.50
Sonnet 5
2026
$31
Sonnet 5.5
2026
$27
Haiku: $ per 1M output tokens (all-in)
Haiku 3.5
2024
$12.40
Haiku 4.5
late 2025
$15.50
Haiku 5.5
2026
$1.55

Falling API prices cut the cost of a fixed amount of work. If a plan's token allowance stayed flat, the dollar value of that allowance at list price fell with it; vendors do not disclose whether allowances changed.

Timeline

  1. Apr 9, 2025 · recalled
    Claude Max launches ($100 for 5x, $200 for 20x)
    Brings Claude Code onto flat-rate plans and sets the $100 / $200 ladder everyone copies.
  2. Aug 28, 2025 · recalled
    Weekly limits added to Pro and Max
    On top of the 5-hour session limit, to curb continuous background use.
  3. Nov 24, 2025 · recalled
    Opus list price falls 3x ($15/$75 to $5/$25)
    Opus 4.5 resets the price of the top Claude tier; later Opus 5.5 trims it to $4/$20.

Sources