Claude Opus 4

anthropicanthropic/claude-opus-4

Claude Opus 4 is Anthropic's most intelligent model and is state-of-the-art for coding and agent capabilities, especially agentic search. It excels for customers needing frontier intelligence: Advanced coding: Independently plan and execute complex development tasks end-to-end. It adapts to your style and maintains high code quality throughout. AI agents: Enable agents to tackle complex, multi-step tasks that require peak accuracy. Agentic search and research: Connect to multiple data sources to synthesize comprehensive insights across repositories. Long-horizon tasks and complex problem solving (virtual collaborator): Unlock new use cases involving long-horizon tasks that require memory, sustained reasoning, and long chains of actions. Content creation: Create human-quality content with natural prose. Produce long-form creative content, technical documentation, marketing copy, and frontend design mockups.

Introducing Claude 4 (announcement) ↗Claude Opus on Amazon Bedrock ↗Anthropic model API overview ↗

Claude Opus 4, released May 22 2025 alongside Sonnet 4, was Anthropic's most powerful model at launch and the coding benchmark leader of its time. It scored 72.5 % on SWE-bench Verified and 43.2 % on Terminal-bench without extended thinking—establishing it as the best coding model available when it shipped. As a hybrid reasoning model it supports both near-instant responses and extended thinking (up to 64 k tokens), and it was the first Anthropic model to show dramatically improved memory capabilities when developers provided local file access.

From a TheRouter routing perspective, Claude Opus 4 occupies the original top of the Anthropic Opus line and is now deprecated. It was superseded by Opus 4.1 (August 2025), which added further agentic precision at the same original price tier, and subsequently by Opus 4.5 through 4.8. Its original pricing was $15/$75 per million tokens, higher than the $5/$25 Opus 4.5+ tier. Existing integrations using the claude-opus-4-20250514 API ID continue to work, but Anthropic recommends migrating to Opus 4.5 or later for better performance and lower cost.

Best for

• Legacy integrations pinned to claude-opus-4-20250514 that have not yet migrated — it remains callable but is deprecated
• Historical benchmarking or research requiring 2025-05-22 vintage model behavior

Reach for something else if

• New production workloads — migrate to anthropic/claude-opus-4-5 or later for lower cost ($5/$25 vs $15/$75), larger context (1 M tokens), and significantly higher coding scores
• Cost-sensitive or high-volume workflows — the original $15/$75 pricing is substantially higher than Opus 4.5+ generations

Context Length

200K

Max Output

32K

Input Priceper 1M tokens

$16.20/ 1M tokens

Output Priceper 1M tokens

$81.00/ 1M tokens

Modalities

textimagepdf→text

Capabilities

Vision

Pricing Breakdown

Type	Rate
Input	$16.20 / 1M tokens
Output	$81.00 / 1M tokens

Supported Parameters

temperaturemax_tokenstop_ptoolstool_choiceresponse_formatreasoningstop

Specifications

Release date	2025-05-22www.anthropic.com ↗	verified
Anthropic API model ID	claude-opus-4-20250514 (alias: claude-opus-4-0)platform.claude.com ↗	verified
AWS Bedrock model ID	anthropic.claude-opus-4-20250514-v1:0platform.claude.com ↗	verified
Vertex AI model ID	claude-opus-4@20250514platform.claude.com ↗	verified
Context window	200 000 tokensplatform.claude.com ↗	verified
Max output tokens	32 000platform.claude.com ↗	verified
Reasoning modes	Instant responses (default) + extended thinking (up to 64 k tokens)www.anthropic.com ↗	verified
Training cutoff	Not publicly disclosed	unknown
License	Anthropic Usage Policy (proprietary, API-only)	verified
Lifecycle status	Deprecated — still callable; migrate to Opus 4.5 or laterplatform.claude.com ↗	verified

Benchmarks

Benchmark	Score	Source
SWE-bench Verified No extended thinking. Led all publicly available models at the time of Claude 4 launch (May 2025).	72.5%%	www.anthropic.com ↗
Terminal-bench No extended thinking. Measures real-world terminal-based software engineering tasks.	43.2%%	www.anthropic.com ↗
GPQA Diamond Without extended thinking; higher when extended thinking is enabled.	74.9%%	www.anthropic.com ↗
MMMLU Without extended thinking.	87.4%%	www.anthropic.com ↗

API Usage Examples

Use the global api.therouter.ai endpoint shown below for new integrations; the legacy China accelerated endpoint is retired.

cURL

curl https://api.therouter.ai/v1/chat/completions   -H "Content-Type: application/json"   -H "Authorization: Bearer $THE_ROUTER_API_KEY"   -d '{
    "model": "anthropic/claude-opus-4",
    "messages": [
      {"role": "user", "content": "Summarize the key points from this input."}
    ]
  }'

API guide

Full API reference →

Chat completion

cURL

curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"anthropic/claude-opus-4","messages":[{"role":"user","content":"Hello"}]}'

More from anthropic

anthropic/claude-opus-4-1Direct drop-in replacement: superior agentic precision at the same original pricing tier.anthropic/claude-opus-4-5Significantly improved coding, agents, and enterprise workflows at $5/$25 — less than a third of Opus 4's original cost.anthropic/claude-sonnet-4Sibling model from the same Claude 4 launch with equivalent SWE-bench score at one-fifth the price.

Similar models

Cross-provider sibling models

openai/o3

OpenAI's frontier reasoning model from the same era; stronger on math, comparable on SWE-bench.

google/gemini-2.5-pro

Google's comparable frontier model from the same period; strong multimodal, competitive coding.

Part of

enterprise governance

Recent coverage

Guide2026-07-23Agentic Coding Model Governance: Operator Guide to Claude Code, Codex, ZCode, and Cursor Enterprise Controls

Frequently asked

Is Claude Opus 4 still available on TheRouter?

Yes. You can call anthropic/claude-opus-4 (or the pinned claude-opus-4-20250514 ID) through TheRouter's OpenAI-compatible endpoint. However, Anthropic has marked it deprecated and recommends migrating to Opus 4.5 or later for better performance and significantly lower cost.

platform.claude.com ↗

How does Claude Opus 4 compare to Sonnet 4 launched at the same time?

Both models were announced on the same day (May 22, 2025) and achieved nearly identical SWE-bench Verified scores (72.5 % vs 72.7 %). Opus 4 led on GPQA Diamond (74.9 % vs 70.0 %) and MMMLU (87.4 % vs 85.4 %), and excelled at long-running agentic tasks. Sonnet 4 offered the same coding benchmark performance at one-fifth the price ($3/$15 vs $15/$75), making it the practical choice for high-volume workloads.

re-authored by TheRouterwww.anthropic.com ↗

Fact ledger — every claim on this page traces here

source	URL	retrieved
Release date	www.anthropic.com ↗	2026-06-10	verified
Anthropic API model ID	platform.claude.com ↗	2026-06-10	verified
AWS Bedrock model ID	platform.claude.com ↗	2026-06-10	verified
Vertex AI model ID	platform.claude.com ↗	2026-06-10	verified
Context window	platform.claude.com ↗	2026-06-10	verified
Max output tokens	platform.claude.com ↗	2026-06-10	verified
Reasoning modes	www.anthropic.com ↗	2026-06-10	verified
Training cutoff	—	—	unknown
License	—	—	verified
Lifecycle status	platform.claude.com ↗	2026-06-10	verified
SWE-bench Verified	www.anthropic.com ↗	2026-06-10	verified
Terminal-bench	www.anthropic.com ↗	2026-06-10	verified
GPQA Diamond	www.anthropic.com ↗	2026-06-10	verified
MMMLU	www.anthropic.com ↗	2026-06-10	verified
Is Claude Opus 4 still available on TheRouter?	platform.claude.com ↗	2026-06-10	to verify
How does Claude Opus 4 compare to Sonnet 4 launched at the same time?	www.anthropic.com ↗	2026-06-10	to verify