Anthropic
Plans
Source: pricing-page · Verified May 3, 2026
Sources
Insights
Computed from plan data, changelog events, and field coverage.
Changelog 10 events
Launched Managed Agents, allowing developers to build AI workers billed per session-hour
API pricing simplified — removed volume-based pricing tiers in favor of flat per-token rates across all usage levels
Boosted usage limits for paid users through the holidays as a temporary goodwill/retention gesture
Added Memory across Conversations feature to the Claude Pro plan
Agent SDK credit breakdown: Team Standard $20/mo, Team Premium $100/mo, Enterprise seat-based Premium $200/mo, Enterprise usage-based $20/mo. Overflow billing toggle off by default; one-time opt-in required. Credits do not pool across teams and do not roll over.
Fast Mode on Opus 4.6 being shut down in ~30 days, driving adoption to Opus 4.8
Opus 4.8 Fast Mode pricing cut by two-thirds vs 4.7: $10/$50 per MTok (2x standard) vs previous $30/$150 (6x). Standard pricing unchanged at $5/$25.
Splitting subscription billing into first-party vs third-party agent/SDK usage effective June 15. New Agent SDK credit pools: Pro $20/mo, Max 5x $100/mo, Max 20x $200/mo. Claude Code weekly limits increased 50% through July 13.
Enterprise seat prices dropped to $20/user/mo (Claude Code) and $10/user/mo (Claude.ai). Seat fee no longer bundles any usage allowance — every token incurs standard API rates on top. Historical 10-15% API discounts removed.
Related Companies
Notes
AI lab (Claude). Hybrid model: per-seat subscriptions for chat product, per-token for API. Enterprise seats dropped to $20/user/mo (Claude Code), $10/user/mo (Claude.ai) in late 2025 — but seat fee no longer bundles any usage. Every token now billed at standard API rates on top. Net effect: lower sticker price, potentially higher TCO for heavy users. API pricing by model: Opus $5/$25/MTok, Sonnet $3/$15, Haiku $1/$5. Prompt caching at 90% off input. Batch at 50%. Web search $10/1K. Managed agents billed per session-hour.