# Anthropic: claude-opus-5

- Model ID: `claude-opus-5`
- Provider: Anthropic
- Web version: https://www.moleapi.com/en/models/anthropic/claude-opus-5
- Content status: deep

## Model introduction

Claude Opus 5 is Anthropic's July 2026 high-capability model for complex agentic coding, code review, enterprise knowledge work, and tasks that require sustained autonomous execution.

## Model capabilities

- OpenAI compatible
- Anthropic compatible
- Prompt cache
- Reasoning

### Verified specifications

- Official positioning: Complex agentic coding and enterprise work
- Release date: 2026-07-24
- Knowledge cutoff: 2026-05
- Context window: 1,000,000 tokens
- Maximum synchronous output: 128,000 tokens
- Input / output: Text, image / Text
- Reasoning levels: low, medium, high, xhigh, max; high by default
- Prompt-cache minimum: 512 tokens
- Verified capabilities: Functions, structured outputs, context editing, memory, code execution, programmatic tools, compaction
- Platform availability: Claude API, AWS Bedrock, Google Vertex AI, Microsoft Foundry

### Independent capability evaluation

- Adaptive reasoning, xhigh effort
- Intelligence: 60 (#2/190)
- Output speed: 52.4 tok/s (#117/190)
- Time to first token: 33.84s
- Source: https://artificialanalysis.ai/models/claude-opus-5-xhigh

The xhigh profile ranks near the top for composite capability, but its measured output speed and first-token delay favor valuable complex work over low-latency interaction.

## Model pricing and access

Pricing is supplied dynamically by the MoleAPI console API.

### Standard (default, x1)

- Input: $5 / 1M tokens
- Output: $25 / 1M tokens
- Cache read: $0.5 / 1M tokens
- Cache write: $6.25 / 1M tokens

### Discount (discount, x0.8)

- Input: $4 / 1M tokens
- Output: $20 / 1M tokens
- Cache read: $0.4 / 1M tokens
- Cache write: $5 / 1M tokens

### Relay (relay, x0.3)

- Input: $1.5 / 1M tokens
- Output: $7.5 / 1M tokens
- Cache read: $0.15 / 1M tokens
- Cache write: $1.875 / 1M tokens

### Access protocols

| Protocol | Method | Endpoint |
| --- | --- | --- |
| anthropic | POST | /v1/messages |
| openai | POST | /v1/chat/completions |

- Live pricing source: https://home.moleapi.com/api/pricing

## About claude-opus-5

Claude Opus 5 is Anthropic’s model for complex agentic coding and enterprise work. Anthropic describes it as a generation that improves on Opus 4.8 in deep reasoning, agentic and long-horizon tasks, and test-time compute scaling; it supports text and image input, text output, and has adaptive thinking enabled by default. [1][2]

## Model Positioning and Core Capabilities

Claude Opus 5 is designed for complex coding and enterprise work that requires execution across files, tools, and multiple stages. Its reasoning depth can be controlled with the `effort` parameter, with adaptive thinking as the default behavior. [1]

Anthropic highlights its capabilities in deep reasoning, long-horizon agentic tasks, and test-time compute scaling. The model provides five reasoning effort levels—`low`, `medium`, `high`, `xhigh`, and `max`—with `high` as the default. [1]

## Capabilities and Specifications

Claude Opus 5 provides a 1 million token context window and a maximum output of 128,000 tokens, supporting text and image input and text output. [1][2]

| Specification | Value |
| --- | --- |
| Context window | 1,000,000 tokens [1] |
| Maximum output | 128,000 tokens [1] |
| Input modalities | Text, images [2] |
| Output modalities | Text [2] |
| Reasoning mode | Adaptive thinking enabled by default [1] |
| Reasoning effort levels | `low`, `medium`, `high`, `xhigh`, `max`; default: `high` [1] |
| Knowledge cutoff | May 2026 [1] |

`max_tokens` constrains the total output budget for both thinking tokens and visible text responses, rather than only limiting the final response length. [1]

The model overview lists Claude Opus 5 as a Claude model with multilingual and vision capabilities. [2]

## Use Cases

Claude Opus 5 covers software engineering tasks such as multi-file feature development, large-scale refactoring, code review, and bug investigation. Anthropic’s prompting guide also uses it for end-to-end task execution, long-context tool use, and multi-agent collaboration. [3]

For enterprise knowledge work, official listed applications include understanding charts, documents, and diagrams; processing complex multi-sheet spreadsheets; and generating presentations. [3]

## Important Limitations

Thinking cannot be disabled for Claude Opus 5 at the `xhigh` or `max` reasoning effort levels; combining either level with `thinking: {"type":"disabled"}` causes the API to return a 400 error. [1]

Disabling thinking applies only to `high` and lower effort levels. Anthropic also notes that when thinking is disabled, the model may occasionally write tool calls into text output rather than `tool_use` blocks, or include internal XML tags in visible responses. [1]

## Sources

[1] [Anthropic Claude Opus 5 更新说明](https://platform.claude.com/docs/en/about-claude/models/whats-new-opus-5)
[2] [Anthropic Claude 模型总览](https://platform.claude.com/docs/en/about-claude/models/overview)
[3] [Claude Opus 5 提示词工程指南](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-5)



## Code examples

### cURL

```bash
curl https://api.moleapi.com/v1/chat/completions \
  -H "Authorization: Bearer $MOLEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-opus-5","messages":[{"role":"user","content":"Hello"}]}'
```

## More related models

- [claude-fable-5](https://www.moleapi.com/en/models/anthropic/claude-fable-5) — Anthropic
- [claude-sonnet-5](https://www.moleapi.com/en/models/anthropic/claude-sonnet-5) — Anthropic
- [claude-opus-4.6](https://www.moleapi.com/en/models/anthropic/claude-opus-4.6) — Anthropic
- [gpt-5.6-luna](https://www.moleapi.com/en/models/openai/gpt-5.6-luna) — OpenAI
- [gpt-5.6-sol](https://www.moleapi.com/en/models/openai/gpt-5.6-sol) — OpenAI
- [gpt-5.6-terra](https://www.moleapi.com/en/models/openai/gpt-5.6-terra) — OpenAI

## Frequently asked questions

### How is claude-opus-5 priced?

Prices are read from the MoleAPI console API and update with model prices, context tiers, and account groups.

### How can I access claude-opus-5?

The catalog currently lists anthropic, openai.

### How do I switch an existing project to claude-opus-5?

Keep the MoleAPI API address and key, replace the model parameter, and check protocol-specific parameters.

### Where do the claude-opus-5 details come from?

Capabilities and limitations are checked against Anthropic and the other cited pages; pricing and protocols come from the MoleAPI console.