Anthropic Makes Claude Sonnet 5 the Default Model for All Free and Pro Users

Anthropic Rolls Out Claude Sonnet 5 Globally

Anthropic launched Claude Sonnet 5 on June 30, 2026, making it the new default model for all Free and Pro plan users on Claude.ai starting July 1.

The model is also available to Max, Team, and Enterprise subscribers, and accessible through Claude Code, the Claude API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.

What Makes Sonnet 5 Different

Claude Sonnet 5 is built for multi-step agentic work.

It can make plans, use tools like web browsers and terminals, and carry tasks through to completion without stopping midway, something earlier Sonnet versions struggled with.

The model also checks its own output without being prompted, which reduces errors in longer automated workflows.

Performance Close to Opus 4.8

Anthropic positions Sonnet 5 as narrowing the gap with its flagship Claude Opus 4.8 model.

On agentic coding benchmarks, Sonnet 5 scores 63.2 percent compared to Opus 4.8 at 69.2 percent and Sonnet 4.6 at 58.1 percent.

On knowledge work tasks, Sonnet 5 slightly outperforms Opus 4.8 while costing considerably less to run.

Pricing and API Access

Introductory pricing runs through August 31, 2026, at $2 per million input tokens and $10 per million output tokens.

From September 1, pricing shifts to $3 per million input tokens and $15 per million output tokens, matching the previous rate of Sonnet 4.6.

Developers can access the model using the API string claude-sonnet-5. Anthropic describes it as a drop-in replacement for Sonnet 4.6 with no need to rebuild existing integrations.

Tokenizer Change Worth Noting

Sonnet 5 uses a new tokenizer that generates between 1 and 1.35 times more tokens than Sonnet 4.6 for the same text input.

Anthropic set the introductory pricing to keep the transition roughly cost-neutral for most users, but high-volume English-language workloads may see a modest cost increase after August.

Safety Improvements

Anthropic says Sonnet 5 shows lower rates of hallucination, sycophancy, and susceptibility to prompt injection compared to Sonnet 4.6.

The model is better at refusing harmful requests and resisting manipulation in automated pipelines, making it more reliable for production deployments.

Industry Context

The Sonnet 5 launch comes as Anthropic competes directly with OpenAI and Google in the agentic AI space.

OpenAI launched GPT-5.6 Sol the week prior, and Google's Gemini 3.5 Flash was similarly positioned as an agentic model built for planning and execution rather than simple conversation.

Anthropic's move to push Sonnet 5 as the universal default signals that mid-tier agentic performance is now the baseline expectation across the industry, not a premium feature.