What Is Claude 3.5 Sonnet?

Anthropic announced the release of Claude 3.5 Sonnet as the first model in the Claude 3.5 series, with access provided at launch via Claude.ai, the Anthropic API, Amazon Bedrock, and Google Cloud Vertex AI.

Subsequently, on October 22, 2024, Anthropic released an upgraded version of Claude 3.5 Sonnet, focusing on improvements to programming, tool use, and computer-operation capabilities. At present, this model mainly exists as a historical reference model: in Anthropic’s deprecation documentation, claude-3-5-sonnet-20240620 and claude-3-5-sonnet-20241022 are listed as being retired on October 28, 2025, and Claude Sonnet 4.6 is recommended as the replacement.

Even though it has been retired, users still search for Claude 3.5 Sonnet because it was widely used in software engineering, document analysis, long-context prompting, image understanding, and agent-like workflow pipelines.

What Are the Main Parameters and Pricing of Claude 3.5 Sonnet?

| Field | Verified Values | |-------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------------------| | Provider | Anthropic (as of June 2026) | | Model series | Claude 3.5 / Claude Sonnet (as of June 2026) | | Model type | Multimodal large language model (as of June 2026) | | Release date | Official announcement date: June 21, 2024; upgraded version released on October 22, 2024 (as of June 2026) | | Context window | 200K tokens, based on Anthropic’s announcement at the time of release (as of June 2026) | | Input pricing | $3.00 per million input tokens; historical release/API pricing (as of June 2026) | | Cached input pricing | As of June 2026, no official channel has confirmed dedicated Claude 3.5 Sonnet–specific pricing | | Output pricing | $15.00 per million output tokens; historical release/API pricing (as of June 2026) | | Billing unit | USD / million tokens / MTok (as of June 2026) | | Supported modalities | Text input, image input, text output (as of June 2026) | | Supported input types | Text and images; upgraded version’s computer-operation beta supports screenshots and tools (as of June 2026) | | Supported output types | Text and code (in text form); native image, audio, or video output has not been confirmed (as of June 2026) | | API access | Retired on October 28, 2025 on the Anthropic API; historically supported: Anthropic API, Amazon Bedrock, Vertex AI (as of June 2026) | | Model ID | claude-3-5-sonnet-20240620 and claude-3-5-sonnet-20241022, both retired (as of June 2026) | | Availability | Claude API retired; Anthropic recommends using claude-sonnet-4-6 instead (as of June 2026) | | Knowledge cutoff | As of June 2026, no official channel has confirmed | | Rate limits | As of June 2026, no official channel has confirmed dedicated limits | | Fine-tuning support | As of June 2026, no official channel has confirmed | | Streaming support | Claude API generally supports streaming output, but it no longer applies after the model is retired (as of June 2026) | | Batch API support | As of June 2026, batch interface support for Claude 3.5 Sonnet has not been confirmed | | Tool/function calling | During its active period, the upgraded Claude 3.5 Sonnet supported tool use and computer-operation beta (as of June 2026) | | Structured output / JSON mode | As of June 2026, no official channel has confirmed dedicated JSON mode | | License/usage restrictions | A proprietary Anthropic model governed by the terms of service; as of June 2026, no model license details have been separately stated |

Note that the above pricing is historical data: at the time of Anthropic’s release, Claude 3.5 Sonnet was priced at $3 per million input tokens and $15 per million output tokens, and it supported a 200K-token context window. Currently, on Anthropic’s deprecation page, the model’s API ID has been marked as retired.

What Practical Capabilities Does Claude 3.5 Sonnet Have in Production?

• Long-Context Document and Code Analysis

  With a 200K-token context window, Claude 3.5 Sonnet can handle long documents and multi-file code snippets, and supports historical and technical specifications. However, it’s important to note that a long context window does not guarantee perfect memory; key outputs still require manual verification.

• Software Engineering Assistance

  Anthropic places emphasis on its capabilities in code generation, code editing, code translation, and multi-step software development workflows. The October 2024 upgraded version improved on the programming and tool-use benchmarks, but production teams still need to run testing, code reviews, and security checks before deployment.

• Vision-Based Understanding Capabilities

  Claude 3.5 Sonnet supports image input, making it suitable for interpreting charts, screenshots, and text in imperfect images. This provides some value in document processing and visual quality inspection, but it cannot replace professional OCR, compliance, or computer-vision systems.

• Agent-Like Computer Operation

  The October 2024 upgraded version introduced a computer-operation public beta. Developers can use screenshots, mouse movement, clicks, and instruction inputs to let Claude interact with the interface. Anthropic has clearly positioned this capability as experimental and prone to errors.

When evaluating Claude 3.5 Sonnet against other production models, teams typically also reference GPT-4o model specifications and API access, Gemini 2.0 Flash multimodal model capabilities, and low-cost alternative options such as GPT-4o mini for scenarios.

What Modalities Does Claude 3.5 Sonnet Support?

| Modality | Supported | Notes | Source status | |-----------------|------------------------------------|------------------------------------------------------------------|------------------------------------------| | Text input | Yes | Standard prompt and dialogue inputs during the API active period | Verified via Anthropic announcements/API documentation | | Image input | Yes | For visual reasoning, charts, screenshots, and text in imperfect images | Verified via Anthropic announcements | | Audio input | Not confirmed | No official channel has confirmed that Claude 3.5 Sonnet supports audio input | Not confirmed as of June 2026 | | Video input | Not confirmed | No official channel has confirmed that Claude 3.5 Sonnet supports video input | Not confirmed as of June 2026 | | Text output | Yes | Main output format | Verified via model behavior and API documentation | | Native image output | Not confirmed | No official materials indicate it is a native image generation model | Not confirmed as of June 2026 | | Audio output | Not confirmed | No official model-dedicated audio output support | Not confirmed as of June 2026 | | Computer interaction | Upgraded beta capability | API tool public beta supports computer operations; experimental and error-prone | From Anthropic’s October 2024 announcement |

What Are the Shortcomings of Claude 3.5 Sonnet?

As of 2026, the most important limitation is availability: Anthropic has marked the Claude 3.5 Sonnet API ID as retired, and new production integrations should not rely on this model. In its deprecation table, Anthropic recommends using Claude Sonnet 4.6 as the replacement.

Claude 3.5 Sonnet also has general AI limitations. For example, hallucinations may occur, key information may be omitted in long contexts, images may be misunderstood, or it may generate code that looks reasonable but is actually incorrect. These are general AI limitations; unless the official documentation explicitly states otherwise, they are not model-specific issues.

In terms of computer operation, Anthropic has explicitly stated that the beta features from October 2024 are not yet fully mature. Operations such as scrolling, dragging, and zooming may be challenging. Teams using agent-like workflow pipelines need to set up sandboxes, permission controls, monitoring, and manual review to ensure the safety of sensitive operations.

According to official materials, Claude 3.5 Sonnet is not a native audio-, video-, or image-generation model. For teams focusing on reasoning or math-intensive tasks, they may consider comparing with specialized reasoning models such as OpenAI o3 or OpenAI o4-mini, but they should be task-oriented rather than relying on overall rankings.

What Scenarios Is Claude 3.5 Sonnet Best Suited For?

| Application scenario | Why it’s suitable | Key limitations | |----------------------------|---------------------------------------------------------------------|----------------------------------------------| | Legacy system operations and maintenance | Suitable for understanding systems that integrated claude-3-5-sonnet-20240620 or claude-3-5-sonnet-20241022 | Retired API IDs should be migrated | | Long-document summarization | A 200K context window supports large-scale prompt inputs | Long-context outputs still require manual verification | | Code review and migration | Anthropic emphasizes code generation, code translation, and multi-step development tasks | Code must be tested and manually reviewed | | Visual document analysis | Image input supports charts, screenshots, and interpretation of non-standard images | Cannot replace professional OCR/compliance systems | | Agent-like workflow research | Computer-operation beta had historical significance in GUI agent experiments | In the evaluated versions, this function is experimental and error-prone |

How Does Claude 3.5 Sonnet Compare With GPT-4o and Gemini 2.0 Flash?

| Comparison dimension | Claude 3.5 Sonnet | GPT-4o | Gemini 2.0 Flash | Scenario fit | |-----------------------|--------------------------------------------------|---------------------------------------------------|---------------------------------------------------|---------------------------------| | Provider | Anthropic | OpenAI | Google | Selection depends on the vendor ecosystem | | Model status in 2026 | Historical legacy; Anthropic API IDs have been retired | Historical legacy; still referenced in many GPT-4o comparisons | Historical Gemini 2.x models; newer versions will follow | Current availability should be confirmed before integration | | Context window | Historically 200K tokens | In the GPT-4o era, APIs commonly use 128K tokens | Google states Gemini 2.0 Flash supports 1M-token context | For long-context needs, new active models can be prioritized | | Supported modalities | Text and image input; text output | Text and vision API support; GPT-4o’s multimodal route is broader | Multimodal input; early 2.0 Flash documentation supported only text output | Choose based on required input/output modalities | | Programming and agent-like workflows | Strong historical performance; upgraded version supports computer-operation beta | Broad support for general multimodal and tool calling | Focused on low-latency multimodal and native tool calling | Recommend testing based on specific tasks | | API status | Anthropic API IDs retired | Before integration, check OpenAI’s current model list | Before integration, check Gemini’s current model list | Avoid hardcoding retired model IDs |

This comparison is limited to verified or widely documented model characteristics and does not make a judgment about which is better or worse. In actual deployment, you should typically prioritize active successor models rather than Claude 3.5 Sonnet itself.

How Can You Access Claude 3.5 Sonnet?

As of June 2026, Claude 3.5 Sonnet should be treated as a retired Anthropic API model and is no longer a target for active integrations. Anthropic has marked claude-3-5-sonnet-20240620 and claude-3-5-sonnet-20241022 as retired on October 28, 2025, and recommends using claude-sonnet-4-6 as a replacement.

During its active period, Claude 3.5 Sonnet could be accessed via Claude.ai, the Anthropic API, Amazon Bedrock, and Google Cloud Vertex AI. Anthropic’s current API overview shows that the Claude API address is the Messages API at POST /v1/messages, and authentication headers such as x-api-key or Authorization, anthropic-version, and content-type are required.

This page does not provide Python or curl runnable examples for Claude 3.5 Sonnet, because all verified model IDs are already retired and those examples no longer apply to the current production environment.

Frequently Asked Questions

What is the context window of Claude 3.5 Sonnet?

According to Anthropic’s release announcement, Claude 3.5 Sonnet supports a 200K-token context window. As of June 2026, this parameter is a historical specification because Anthropic has marked the related API IDs as retired.

How much does Claude 3.5 Sonnet cost?

When Anthropic released it, the pricing was $3 per million input tokens and $15 per million output tokens. Since the API ID is retired, current production pricing no longer applies.

Can developers still access Claude 3.5 Sonnet via the Anthropic API?

Anthropic’s deprecation documentation has marked claude-3-5-sonnet-20240620 and claude-3-5-sonnet-20241022 as retired on October 28, 2025. New integrations should choose an active replacement model.

What are the common application scenarios for Claude 3.5 Sonnet?

Claude 3.5 Sonnet is mainly used for programming assistance, document analysis, long-context reasoning, visual understanding, and early computer-operation workflow pipelines. At present, these uses are mainly for legacy migrations and historical comparisons.