Anthropic Makes Claude Sonnet 5 Pricing Permanent and Enables Auto Mode by Default Ahead of IPO

Anthropic is accelerating its commercialization strategy ahead of an expected autumn 2026 IPO by permanently locking in Claude Sonnet 5's introductory pricing at $2 per million input and $10 per million output tokens, canceling a planned 50% price hike. Simultaneously, Anthropic is enabling auto mode by default in Claude Code starting August 14, 2026, absorbing the associated classifier token costs. This shift is backed by data showing human testers fail to block 86.4% of dangerous commands, whereas auto mode blocks 89%.

Claude Sonnet 5 permanent pricing

• ▪Anthropic announced on August 10, 2026, that the introductory API pricing for the Claude Sonnet 5 model will be permanently retained at $2 per million input tokens and $10 per million output tokens.
• ▪The decision to lock in Claude Sonnet 5 pricing permanently cancels a planned price increase to $3 per million input tokens and $15 per million output tokens that was set to take effect on September 1, 2026.

Sonnet 5 agentic capabilities

• ▪In pre-deployment safety evaluations, Claude Sonnet 5 demonstrated a lower rate of undesirable behaviors, better refusal of malicious requests, and stronger resistance to prompt injection attacks than Claude Sonnet 4.6.
• ▪Claude Sonnet 5 is launched with cyber safeguards enabled by default, which are the same real-time safeguards present in Claude Opus 4.7 and 4.8 to detect and block dangerous cyber usage.
• ▪Claude Sonnet 5 is designed to autonomously formulate plans, use tools like browsers and terminals, and run autonomously at a level that previously required larger, more expensive models.

Claude Code auto mode

• ▪To secure auto mode, Anthropic added rules classifying data exfiltration as a hard deny, checking repository publicity before git push operations, and scanning for injection attempts via API-side probes.
• ▪Anthropic announced that starting August 14, 2026, Claude Code's auto mode will be enabled by default for Pro, Max, and Team subscribers, while Enterprise users temporarily retain manual options.
• ▪Anthropic will absorb the additional token costs incurred by related classifiers in Claude Code's automatic mode, meaning users will no longer be charged for this overhead.

Human approval ineffectiveness data

• ▪Anthropic's data shows that users approve 97% of individual Claude Code permission prompts, whereas they reject 39% of complete plans submitted by Claude for approval
• ▪In adversarial testing conducted with Apollo Research, Claude Code's auto mode attack miss rate was reduced from 12% to 7% after Anthropic strengthened the classifier with additional environmental context.
• ▪In a controlled experiment with 1,053 professional testers, human testers blocked a dangerous command in only 13.6% of cases, whereas Claude Code's auto mode blocked the same command in 89% of cases.

Pre-IPO commercialization strategy

• ▪OpenAI also adjusted its commercial strategy on August 10, 2026, by launching premium seat options for ChatGPT Business priced at $125 per user per month, targeting heavy users.
• ▪Anthropic confidentially filed its IPO registration documents with the U.S. Securities and Exchange Commission on June 1, 2026, with a market expectation of listing in autumn 2026.
• ▪As of early May 2026, Anthropic's annualized revenue surpassed $47 billion, following the completion of a $65 billion Series H funding round that valued the company at $965 billion.