Claude Sonnet 4 balances impressive performance for coding with the right speed and cost for high-volume use cases: Coding: Handle everyday development tasks with enhanced performance-power code reviews, bug fixes, API integrations, and feature development with immediate feedback loops. AI Assistants: Power production-ready assistants for real-time applications—from customer support automation to operational workflows that require both intelligence and speed. Efficient research: Perform focused analysis across multiple data sources while maintaining fast response times. Ideal for rapid business intelligence, competitive analysis, and real-time decision support. Large-scale content: Generate and analyze content at scale with improved quality. Create customer communications, analyze user feedback, and produce marketing materials with the right balance of quality and throughput.
Claude Sonnet 4, released May 22 2025, is Anthropic's mid-tier hybrid reasoning model designed for high-volume production workloads. It delivers frontier-level coding (SWE-bench Verified 72.7 %, 80.2 % with high compute) and balanced reasoning at $3/$15 per million tokens, making it a practical drop-in upgrade from Sonnet 3.7 for everyday development, agentic sub-tasks, and real-time customer-facing applications.
From a TheRouter operator perspective the key operational features are the hybrid instant + extended-thinking modes, the 1 M context window (beta, retiring April 30 2026), and strong instruction-following that makes it reliable as a sub-agent within multi-agent systems. Sonnet 4 is succeeded by Sonnet 4.5 (late 2025) and Sonnet 4.6 (Feb 2026), both of which retain the same price tier while adding deeper long-context and computer-use capabilities.
Anthropic's lifecycle table lists the native Claude API id 'claude-sonnet-4-20250514' as retired on Anthropic-operated platforms after a 2026-04-14 deprecation notice and 2026-06-15 retirement date. Anthropic recommends migrating to 'claude-sonnet-4-6'.
TheRouter still keeps the public model page and OpenAI-compatible id 'anthropic/claude-sonnet-4' for customers who need a stable reference, but new integrations should route to 'anthropic/claude-sonnet-4.6' because the upstream retirement can affect live availability and support guarantees.
| Type | Rate |
|---|---|
| Input | $3.24 / 1M tokens |
| Output | $16.20 / 1M tokens |
| Release date | 2025-05-22www.anthropic.com ↗ | verified |
| Anthropic model id | claude-sonnet-4-20250514aws.amazon.com ↗ | verified |
| Training cutoff | Not publicly disclosed | unknown |
| Context window | 200 k standard; 1 M via context-1m-2025-08-07 beta (retiring 2026-04-30)platform.claude.com ↗ | verified |
| Max output tokens | 64 000www.anthropic.com ↗ | verified |
| Reasoning modes | Instant responses (default) + extended thinking (up to 64 k tokens)www.anthropic.com ↗ | verified |
| Deployment | Anthropic API, Amazon Bedrock, Google Cloud Vertex AIaws.amazon.com ↗ | verified |
| License | Anthropic Usage Policy (proprietary, API-only) | verified |
| Benchmark | Distribution | Score | Source |
|---|---|---|---|
SWE-bench Verified No extended thinking; high-compute parallel sampling reaches 80.2 %. | 72.7%% | www.anthropic.com ↗ | |
GPQA Diamond Without extended thinking; higher with extended thinking enabled. | 70.0%% | www.anthropic.com ↗ | |
MMMLU Without extended thinking. | 85.4%% | www.anthropic.com ↗ | |
MMMU Without extended thinking. | 72.6%% | www.anthropic.com ↗ | |
AIME Without extended thinking. | 33.1%% | www.anthropic.com ↗ | |
OSWorld Computer-use benchmark; Sonnet 4.5 later improved this to 61.4 %. | 42.2%% | www.anthropic.com ↗ |
Use the global api.therouter.ai endpoint shown below for new integrations; the legacy China accelerated endpoint is retired.
curl https://api.therouter.ai/v1/chat/completions -H "Content-Type: application/json" -H "Authorization: Bearer $THE_ROUTER_API_KEY" -d '{
"model": "anthropic/claude-sonnet-4",
"messages": [
{"role": "user", "content": "Summarize the key points from this input."}
]
}'curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"anthropic/claude-sonnet-4","messages":[{"role":"user","content":"Hello"}]}'Anthropic's deprecation table says Claude Sonnet 4 was deprecated on 2026-04-14 and retired on 2026-06-15, with Claude Sonnet 4.6 named as the replacement. Treat TheRouter's page as a compatibility reference, not a recommendation for new deployments.
Anthropic introduced Sonnet 4.5 with stronger coding, computer-use, and long-horizon agent performance while keeping Sonnet 4's input/output price. It became the practical upgrade path before Sonnet 4.6 was named the retirement replacement.
No. Anthropic lists the native Sonnet 4 API model as retired and recommends Claude Sonnet 4.6. Keep Sonnet 4 only for compatibility checks, historical comparisons, or controlled migrations where you have already validated live upstream availability.
The launch facts remain useful for capability history: Sonnet 4 shipped with hybrid instant plus extended thinking modes, $3/$15 pricing, 64k max output, and 72.7 % SWE-bench Verified. The lifecycle changed later: Anthropic retired the native API model and pointed users to Sonnet 4.6.
No. TheRouter can normalize request and response shape, model ids, and routing behavior, but it cannot turn a retired Anthropic-operated upstream into an active Anthropic model. Use the divergence note on this page to separate TheRouter compatibility from vendor lifecycle support.
| source | URL | retrieved | |
|---|---|---|---|
| Release date | www.anthropic.com ↗ | 2026-05-25 | verified |
| Anthropic model id | aws.amazon.com ↗ | 2026-05-25 | verified |
| Training cutoff | — | — | unknown |
| Context window | platform.claude.com ↗ | 2026-09-10 | verified |
| Max output tokens | www.anthropic.com ↗ | 2026-05-25 | verified |
| Reasoning modes | www.anthropic.com ↗ | 2026-05-25 | verified |
| Deployment | aws.amazon.com ↗ | 2026-05-25 | verified |
| License | — | — | verified |
| SWE-bench Verified | www.anthropic.com ↗ | 2026-05-25 | verified |
| GPQA Diamond | www.anthropic.com ↗ | 2026-05-25 | verified |
| MMMLU | www.anthropic.com ↗ | 2026-05-25 | verified |
| MMMU | www.anthropic.com ↗ | 2026-05-25 | verified |
| AIME | www.anthropic.com ↗ | 2026-05-25 | verified |
| OSWorld | www.anthropic.com ↗ | 2026-05-25 | verified |
| Anthropic deprecates Claude Sonnet 4 on the Claude API | platform.claude.com/model-deprecations ↗ | 2026-09-10 | verified |
| Claude Sonnet 4.5 released at the same $3/$15 Sonnet price tier | anthropic.com/news ↗ | 2026-09-10 | verified |
| Should I start a new TheRouter integration on Claude Sonnet 4? | platform.claude.com ↗ | 2026-09-10 | to verify |
| What changed since the original Sonnet 4 launch page? | www.anthropic.com ↗ | 2026-09-10 | to verify |
| Does TheRouter's OpenAI-compatible API surface change the Anthropic retirement status? | platform.claude.com ↗ | 2026-09-10 | to verify |