Overview
Claude Opus 4 is a large language model from Anthropic, the flagship of the Claude 4 generation released on May 22, 2025. At launch Anthropic positioned Claude Opus 4 as its most capable model and described it as the world's best coding model, built for complex, long-running tasks that can require thousands of steps and several hours of continuous work.
Claude Opus 4 accepts text and image input and returns text, with a 200K-token context window and up to 32K output tokens. It is a hybrid reasoning model that offers both near-instant responses and an extended-thinking mode for deeper, step-by-step reasoning. Pricing at launch was $15 per million input tokens and $75 per million output tokens.
Claude Opus 4 was superseded by Claude Opus 4.1 (August 2025) and later Opus 4.x releases, and the original snapshot (claude-opus-4-20250514) has since been retired. It remains a notable milestone as the model that launched Anthropic's Claude 4 line and its leading position on agentic coding benchmarks.
| Released | 2025-05-22 |
|---|---|
| License | Proprietary |
| Weights | API only |
| Context | 200K |
| Max output | 32K |
| Modalities | Text, Vision |
| Status | Retired |
Benchmarks
- SWE-bench Verified72.5%
- Terminal-bench43.2%
- GPQA Diamond74.9%
- MMLU87.4%
- MMMU73.7%
- AIME 202533.9%
Scores on a 0–100 scale (25-point gridlines); higher is better. Each benchmark links to its published source.
Pricing
| Input | $15.00 per million tokens |
|---|---|
| Output | $75.00 per million tokens |
Strengths
- Agentic coding — 72.5% on SWE-bench Verified at launch, which Anthropic cited as best-in-class
- Sustained long-horizon work — designed to run continuously for hours across thousands of steps
- Hybrid reasoning with an extended-thinking mode for harder, multi-step problems
- Strong graduate-level reasoning and knowledge (74.9% GPQA Diamond, 87.4% MMLU)
- Multimodal text + image understanding (73.7% MMMU)
Best for
- Autonomous and agentic coding workflows
- Large-codebase refactors and multi-file software tasks
- Long-running agents that execute many sequential steps
- Complex reasoning and technical research
- Document and image analysis
How to access
| Provider | Model ID |
|---|---|
| Anthropic ↗ | claude-opus-4-20250514 |
Claude Opus — every version
The full lineage of the Claude Opus line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.
| Version | Released | Context | License |
|---|---|---|---|
| Claude Opus 4.8current | 2026-05-28 | 1M | Proprietary |
| Claude Opus 4.7 | 2026-04-16 | — | Proprietary |
| Claude Opus 4.6 | 2026-02-05 | — | Proprietary |
| Claude Opus 4.5 | 2025-11-24 | — | Proprietary |
| Claude Opus 4.1 | 2025-08-05 | — | Proprietary |
| Claude Opus 4 | 2025-05-22 | — | Proprietary |
| Claude 3 Opus | 2024-03-04 | — | Proprietary |
FAQ
When was Claude Opus 4 released?
Anthropic released Claude Opus 4 on May 22, 2025, as the flagship model of the Claude 4 generation alongside Claude Sonnet 4.
What is Claude Opus 4's context window and max output?
Claude Opus 4 has a 200K-token context window and can generate up to 32K output tokens per response. It accepts both text and image input.
How much did Claude Opus 4 cost?
Claude Opus 4 was priced at $15 per million input tokens and $75 per million output tokens via the Anthropic API.
Is Claude Opus 4 still available?
No. The original snapshot (claude-opus-4-20250514) has been retired. It was succeeded by Claude Opus 4.1 in August 2025 and later Opus 4.x models; Anthropic recommends migrating to the current Claude Opus tier.