Claude Sonnet 4 is Anthropic’s mid-size model, described as a hybrid reasoning model with two modes: near-instant responses and extended thinking for deeper reasoning. It is presented as a balanced model with strong coding and reasoning capabilities.
Anthropic says Claude Sonnet 4 improves on Claude Sonnet 3.7 across multiple areas, especially coding. It is also described as offering greater control over how eagerly it implements changes.
| Developer | Anthropic |
|---|---|
| Model ID | anthropic/claude-sonnet-4 |
| Release Date | May 21, 2025 |
| Context Window | 1,000,000 tokens (~1M) |
| Max Output Tokens | 64,000 tokens |
| Modalities | Image, Text → Text |
| Capabilities | file input, parallel tool calls, reasoning, streaming, structured (JSON schema) output, tool/function calling, vision, web search |
Hybrid reasoning
Uses two modes: near-instant responses and extended thinking for deeper reasoning.
Balanced model
Described as Anthropic’s mid-size balanced model.
Coding focus
Anthropic highlights stronger coding performance and broad coding improvements versus Claude Sonnet 3.7.
Reasoning and instruction following
AWS describes it as having strong coding and reasoning capabilities with improved instruction following.
Tool use support
Anthropic notes that the model can alternate between reasoning and tool use, including web search, to improve responses.
Coding
Recommended for coding workloads and code-related assistance.
In-depth research
Positioned for in-depth research tasks.
Agents
Presented as suitable for agentic workflows and multi-step tool use.
High-volume use
Google Cloud describes it as suitable for high-volume use cases.
Claude Sonnet 3.7
Anthropic describes Claude Sonnet 4 as an improvement over Claude Sonnet 3.7, especially for coding.
Claude Opus 4
Anthropic groups Claude Sonnet 4 with Claude Opus 4 as part of the Claude 4 generation of hybrid reasoning models.
Sources: https://www.anthropic.com/news/claude-4 · https://www.anthropic.com/claude/sonnet · https://www.anthropic.com/research/claude-4
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
Claude Sonnet 4 is documented as suitable for customer-facing AI assistants and high-volume tasks. It is also suited to coding work, including code reviews, bug fixes, and broader development workflows.
For larger-scale software work, it is documented as useful for agentic coding and multi-step development tasks, including planning, writing, debugging, and large refactors. It is also suited to AI assistants and near real-time production workloads, where responsiveness and high throughput matter.
Claude Sonnet 4 is documented as useful for knowledge Q&A over large documents, codebases, or datasets. It is also suited to large-scale content generation and analysis, including efficient research and summarization of extensive material.
It is documented as useful for agent-based workflows that maintain context across many tool calls and multi-step processes.
import requests r = requests.post( "https://api.aimlapi.com/v1/chat/completions", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, json={ "model": "anthropic/claude-sonnet-4", "messages": [ { "role": "user", "content": "Hello!" } ] }, ) print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", { method: "POST", headers: { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }, body: JSON.stringify({ "model": "anthropic/claude-sonnet-4", "messages": [ { "role": "user", "content": "Hello!" } ] }), }); console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"anthropic/claude-sonnet-4","messages":[{"role":"user","content":"Hello!"}]}'
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Input | |
| Output | |
| Cached input |
Claude Sonnet 4 documents cache write tokens as billed at 1.25× the base input price for 5-minute caching and 2× for 1-hour caching.
| Benchmark | Score | What it measures | Source | Retrieved |
|---|---|---|---|---|
| Intelligence | 18.9 | Composite score across standardised reasoning, knowledge and problem-solving evaluations, measured independently by Artificial Analysis | Source | September 22, 2026 |
| Coding | 37.6 | Composite score across standardised coding evaluations, measured independently by Artificial Analysis | Source | September 22, 2026 |
| Math | 74.3 | Composite score across standardised mathematics evaluations, measured independently by Artificial Analysis | Source | September 22, 2026 |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
Claude Sonnet 4 This page | Agentic workflows and structured output | |||
| Complex agentic coding and enterprise workflows | ||||
| Chat + assistants | ||||
| Chat + assistants | ||||
| Multimodal reasoning |
Claude Sonnet 4 has a 1,000,000 tokens context window and can return up to 64,000 tokens.
Claude Sonnet 4 takes image, text as input and returns text.
Use anthropic/claude-sonnet-4 as the model id. Requests go to https://api.aimlapi.com/v1/chat/completions.
Claude Sonnet 4 became available on May 22, 2025.
Claude Sonnet 4 is priced at input $4.1262 / 1M tokens, output $20.631 / 1M tokens, cached input $0.41262 / 1M tokens.
Yes, Claude Sonnet 4 can stream responses as they are generated.
Yes, Claude Sonnet 4 accepts image input alongside text.
Claude Sonnet 4 was built by Anthropic.
Yes, Claude Sonnet 4 accepts image and text input alongside text output, and supports vision as a listed capability.
Yes, it supports tool use, including parallel tool calls.
