Anthropic Launches Claude 3.7 Sonnet Amid Infrastructure Struggles and Pricing Controversies

Mar 13, 2026

Anthropic has launched Claude 3.7 Sonnet, its first hybrid reasoning model, priced at $3/$15 per million tokens. The launch comes as the company struggles with infrastructure, suffering ten service disruptions since June 5, and after pausing controversial token-based billing changes for its Agent SDK following developer backlash. A new agentic coding tool, Claude Code, was also released in a limited preview.

Claude 3.7 Sonnet Launch and Features

▪Anthropic announced Claude 3.7 Sonnet, its first hybrid reasoning model.
▪The extended thinking mode in Claude 3.7 Sonnet is available on all Claude plans except for the free tier.
▪Claude 3.7 Sonnet can produce near-instant responses or use an "extended thinking" mode for step-by-step reasoning.
▪Claude 3.7 Sonnet is available on all Claude plans, the Claude Developer Platform, Amazon Bedrock, and Google Cloud’s Vertex AI.

Infrastructure and Reliability Issues

▪Anthropic's Claude AI platform experienced its tenth significant service disruption since June 5, 2026.
▪A June 16, 2026 outage affected Anthropic's Claude Opus 4.8 and Haiku 4.5 models.
▪Anthropic has not published post-incident root cause analyses for its service disruptions in June 2026.
▪Anthropic's annualized revenue grew from $9 billion at the end of 2025 to over $30 billion in early April 2026.

Model Pricing and Billing Changes

▪Claude 3.7 Sonnet is priced at $3 per million input tokens and $15 per million output tokens, the same as its predecessors.
▪Anthropic paused planned billing changes for its Claude Agent SDK after developers warned of a major cost increase.
▪Claude Opus 4.8 is priced at $5 per million input tokens and $25 per million output tokens.
▪A proposed class-action lawsuit filed June 14, 2026, alleges Anthropic's Claude Max plans provide less usage than advertised.

Developer Tools and Ecosystem

▪Claude Code can search and read code, edit files, run tests, and commit code to GitHub.
▪Anthropic introduced Claude Code, a command line tool for agentic coding, in a limited research preview.

Model Capabilities and Performance

▪According to a February 2026 report, Claude Code accounts for roughly 4% of all public GitHub commits.
▪In a June 2026 evaluation by YZ Index, Claude Sonnet 4.6's code execution score dropped from 100 to 50 in a single day.
▪Claude 3.7 Sonnet achieves state-of-the-art performance on the SWE-bench Verified benchmark, which evaluates solving real-world software issues.

Safety and Data Policies

▪The US government issued an export control directive to suspend all access to Anthropic's Fable 5 and Mythos 5 models.
▪Claude 3.7 Sonnet reduces unnecessary refusals by 45% compared to its predecessor.
▪Claude Fable 5 requires 30-day data retention and is not available under zero data retention (ZDR) arrangements.