Stable existing integrations
Keep a pinned production model while testing Sonnet 5 for regressions, quality, latency, and cost.
Independent tool overview
Claude Sonnet 4.6 at a glance
Claude Sonnet 4.6 is an active legacy Anthropic model with 1M context, adaptive thinking, vision, coding, and tool use, but Claude Sonnet 5 is now faster-moving, stronger, and cheaper.
Visit the official Claude Sonnet 4.6 site ↗
- Developer
- Anthropic
- Status
- Active legacy model
- Model ID
- claude-sonnet-4-6
- Context
- 1 million tokens
- Recommended successor
- Claude Sonnet 5
Overview
What Claude Sonnet 4.6 is
Anthropic released Claude Sonnet 4.6 in February 2026 as a balanced model for coding, agents, search, analysis, and everyday knowledge work. It supports text and image input, a 1-million-token context window, up to 128K output, adaptive thinking, and tool use.
The model remains available through the Claude API and major cloud platforms, but Anthropic now labels it legacy and recommends migrating to Claude Sonnet 5. Sonnet 4.6 has a fixed model ID, so it does not silently become the newer Sonnet model.
Existing production systems can keep using Sonnet 4.6 while they validate a migration, and Anthropic says retirement will be no sooner than February 17, 2027. New integrations should normally start with Sonnet 5 because it improves agentic performance and now costs less per token.
Use cases
Who Claude Sonnet 4.6 is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Long-context analysis
Process large document sets, codebases, and mixed text-image inputs within a 1M-token window.
Agentic coding
Use tools, code execution, web capabilities, and adaptive thinking for multi-step software work.
Batch transition work
Run lower-priority processing at a 50% Batch API discount during a controlled migration period.
Capabilities
Core Claude Sonnet 4.6 features
One-million-token context
The full 1M context window is available at standard per-token rates without a special long-context surcharge.
Adaptive thinking
The model can vary its reasoning, with configurable effort and high as the current default.
Text and vision
Sonnet 4.6 accepts text and images and returns text for analysis, coding, and document workflows.
Large outputs
Current model documentation lists up to 128K output tokens, with a larger beta maximum for Batch API jobs.
Tool ecosystem
The model supports tool use and Anthropic server-side tools for search, fetching, code execution, memory, and related workflows.
Cloud availability
It remains available through the Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry, and Claude Platform on AWS.
Process
How the Claude Sonnet 4.6 workflow works
Step 1
Identify dependencies
Inventory prompts, tool schemas, effort settings, evaluation thresholds, and any behavior tied specifically to Sonnet 4.6.
Step 2
Build a migration set
Create representative coding, agent, analysis, safety, latency, and cost cases from real production traffic.
Step 3
Compare Sonnet 5
Run the same set against claude-sonnet-5 and inspect task success, tool use, output format, and token consumption.
Step 4
Fix breaking assumptions
Remove unsupported assistant prefilling and use structured outputs or explicit prompt instructions instead.
Step 5
Roll out gradually
Canary the successor, monitor errors and business outcomes, and preserve a rollback path until results are stable.
Cost
Claude Sonnet 4.6 pricing and free plan
Claude Sonnet 4.6 remains priced at $3 per million input tokens and $15 per million output tokens. Prompt caching and Batch API discounts can reduce repeat and asynchronous workload costs.
Standard input
$3 / 1M tokens
Input pricing for standard Sonnet 4.6 API requests.
- 1M context included
- Global inference
- Usage based
Standard output
$15 / 1M tokens
Generated-token pricing, including visible answer and supported reasoning usage.
- Up to 128K output
- Effort affects token use
- Usage based
Prompt caching
$3.75 5m write / $6 1h write / $0.30 read per 1M
Cache pricing for repeatedly reused prompt and context content.
- 5-minute or 1-hour writes
- Lower read price
- Useful for repeated long context
Batch API
50% off input and output
Discounted asynchronous processing for workloads that do not need immediate responses.
- Asynchronous
- Separate operational constraints
- Migration and evaluation friendly
Claude Sonnet 5
$2 input / $10 output per 1M
Current successor pricing, now permanently below Sonnet 4.6.
- Recommended for new integrations
- 1M context
- Stronger agentic performance
Pricing checked . Check current pricing at the source ↗
Assessment
Claude Sonnet 4.6 strengths and limitations
Where it stands out
- Stable pinned model ID for reproducible existing integrations
- One-million-token context with no separate long-context price multiplier
- Strong coding, analysis, vision, and tool-use capabilities
- Adaptive thinking and effort controls help balance quality and latency
- Broad availability across first-party and major cloud platforms
What to consider
- Anthropic classifies Sonnet 4.6 as legacy and explicitly recommends Sonnet 5
- Sonnet 5 is stronger on agentic work and permanently cheaper at current prices
- Reliable knowledge is concentrated through August 2025, so current facts require tools or retrieval
- Assistant-message prefilling is not supported and returns an error
- The dateless Sonnet 4.6 ID is a fixed snapshot, not an alias that upgrades to newer models
- Continued support is time-limited; retirement is committed only through at least February 17, 2027
Compare
Claude Sonnet 4.6 alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Consumer
Claude Sonnet 5
The direct recommended successor with better agentic performance and lower token prices.
Explore Claude Sonnet 5 →Consumer
Claude Opus 4.8
A stronger legacy-era Claude option for demanding reasoning and agent workflows, though at higher cost.
Explore Claude Opus 4.8 →Business Operations
DeepSeek
A much lower-cost open-weight alternative with 1M context and modern agent APIs.
Explore DeepSeek →Business Operations
ChatGPT
A broader managed assistant and platform for teams that want a first-party user interface plus developer options.
Explore ChatGPT →Questions
Claude Sonnet 4.6 FAQs
Is Claude Sonnet 4.6 still available?
Yes. Anthropic lists it as active but legacy across the Claude API and supported cloud platforms.
What replaced Claude Sonnet 4.6?
Claude Sonnet 5 is the recommended successor. Anthropic says it improves reasoning, tool use, coding, knowledge work, and agentic performance.
When will Claude Sonnet 4.6 retire?
Anthropic states that retirement will be no sooner than February 17, 2027. Teams should monitor the official deprecation schedule rather than treating that as a permanent support date.
How much does Claude Sonnet 4.6 cost?
The standard Claude API price is $3 per million input tokens and $15 per million output tokens. Prompt caching has separate write/read prices, and Batch API jobs receive a 50% discount.
Does claude-sonnet-4-6 automatically point to Sonnet 5?
No. Anthropic says dateless IDs from the 4.6 generation onward identify fixed model snapshots. Migrating to Sonnet 5 requires changing the model ID.
Should a new project use Sonnet 4.6?
Usually no. Start with Sonnet 5 unless a specific compatibility or validation requirement demands Sonnet 4.6, because the successor is both recommended and cheaper.
Bottom line
Our Claude Sonnet 4.6 verdict
Claude Sonnet 4.6 remains a capable and stable legacy model for existing long-context, coding, and agent integrations. It is no longer the sensible default for new work: Sonnet 5 is the official successor, improves the most important agentic capabilities, and has lower permanent API prices. Use Sonnet 4.6 as a controlled migration baseline, not a destination.
Visit Claude Sonnet 4.6 website ↗Similar Consumer Tools
View all Consumer AI tools →
Qwen3-TTS CustomVoice 1.7B
Qwen3-TTS CustomVoice 1.7B - Alibaba's multilingual text-to-speech model with voice cloning using just three seconds of reference audio
Lyria 3
Lyria 3 - Google's new AI music generation model for creating 30-second, customizable audio clips from text, image, or video prompts.
Tiny Aya
Tiny Aya - Cohere Labs' new open-source multilingual small model covering 70+ languages in just 3.35B parameters
Gemini 3.1 Pro
Gemini 3.1 Pro - Google's upgraded flagship model with SOTA reasoning gains
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.