claude-4-sonnet

claude-4-sonnet

by Anthropic
3.0
Visit Website
LinkedIn X (Twitter) YouTube

Pros & Cons

Pros

  • Large Context and Low Hallucination: A 200K-token context window and comparatively low hallucination rate suited reviewing long contracts and case files.
  • Strong Agentic and Coding Performance: Selected to power GitHub Copilot's coding agent at launch, reflecting strong multi-step task execution.
  • Hybrid Reasoning Modes: Could switch between fast responses and extended step-by-step thinking depending on task complexity.

Cons

  • Fully Retired: Anthropic retired Claude Sonnet 4 on June 15, 2026; API calls now fail with no automatic failover, making it unusable for current workflows.
  • No Legal-Specific Training: Not trained on specialized legal datasets, so it lacked jurisdiction-specific legal reasoning and citation grounding.
  • Requires Active Migration: Because there is no automatic failover, any tool still referencing this model ID needs code changes to migrate to Claude Sonnet 4.5/4.6.

About claude-4-sonnet

Introduction

Claude Sonnet 4 was Anthropic's hybrid reasoning model released May 22, 2025, offering near-instant responses or extended step-by-step thinking, a 200K-token context window, up to 64K output tokens, and strong agentic coding performance (it powered GitHub Copilot's coding agent at launch). Like other general-purpose Anthropic models, it was not trained on specialized legal datasets, so it lacked deep jurisdiction-specific legal reasoning compared with purpose-built legal AI tools, despite its Constitutional AI safety foundation.

2026 Status: Retired

Anthropic retired Claude Sonnet 4 on June 15, 2026, following an April 2026 deprecation notice. API calls to the claude-sonnet-4 model ID now fail outright, with no automatic failover to a newer model, so any application still pointing at it needs code changes, not just time, to keep working. Anthropic's recommended successor is Claude Sonnet 4.5 (with Claude Sonnet 4.6 also available as a further update); Claude Opus 4.6/4.8 are alternatives for more complex or agentic workloads. Firms using Claude Sonnet 4 in any legal tooling should treat migration as urgent rather than optional.

Use in Legal Work

While available, Claude Sonnet 4's large context window and low hallucination rate made it a reasonable general-purpose choice for reviewing lengthy contracts or summarizing case files, and its agentic capabilities suited multi-step drafting workflows. It never had native access to legal databases or built-in citation verification, so legal use always required professional review — and with the model now fully retired and returning hard errors, it is not usable for any current or new legal-AI workflow.

Frequently Asked Questions

No. Anthropic retired it on June 15, 2026, and API calls to the claude-sonnet-4 model ID now fail with no automatic failover. Anthropic's recommended successor is Claude Sonnet 4.5 (or the newer Claude Sonnet 4.6).

Its large 200K-token context window and low hallucination rate made it a solid general-purpose choice for summarizing or cross-referencing long contracts and case files, but it had no legal-database grounding or citation verification, so outputs still required review by a qualified legal professional.

Reviews