Anthropic Releases Claude 3.5 Sonnet With Major Coding Upgrades

Anthropic has released Claude 3.5 Sonnet, the first model in its upcoming Claude 3.5 family. The new model outperforms Claude 3 Opus and competing models across a wide range of benchmarks while operating at twice the speed and the cost of the mid-tier Claude 3 Sonnet, according to the company.

The release date was August 28, 2025.

Coding capabilities

Claude 3.5 Sonnet achieved 64% accuracy on Anthropic's internal agentic coding evaluation, compared to 38% for Claude 3 Opus. The evaluation tests the model's ability to fix bugs or add features to open-source codebases using natural language descriptions. When given appropriate tools, the model can independently write, edit, and execute code with "sophisticated reasoning and troubleshooting capabilities," according to the announcement.

The model also shows strong performance on graduate-level reasoning (GPQA), undergraduate-level knowledge (MMLU), and coding proficiency (HumanEval), setting new industry benchmarks in each category.

Vision improvements

Claude 3.5 Sonnet is Anthropic's strongest vision model yet, surpassing Claude 3 Opus on standard vision benchmarks. The improvements are most noticeable for tasks requiring visual reasoning, such as interpreting charts and graphs. The model can also accurately transcribe text from imperfect images—a capability useful for retail, logistics, and financial services.

Pricing and availability

The model is available for free on Claude.ai and the Claude iOS app. Claude Pro and Team subscribers get significantly higher rate limits. API pricing is $3 per million input tokens and $15 per million output tokens, with a 200K token context window. It is also available via Amazon Bedrock and Google Cloud Vertex AI.

Safety and what's next

Anthropic says red teaming assessments conclude Claude 3.5 Sonnet remains at ASL-2 (the company's safety level for models without catastrophic misuse potential). The model was also evaluated by the UK AI Safety Institute before release, with results shared with the US AI Safety Institute.

The company plans to release Claude 3.5 Haiku and Claude 3.5 Opus later this year to complete the 3.5 model family.

Anthropic also introduced "Artifacts" on Claude.ai—a new feature that lets users generate code snippets, documents, or website designs in a dedicated side panel alongside their conversation.