Loading model…正在加载模型详情…
Chat with Claude Opus 4.8, Anthropic’s model for complex reasoning and sustained coding work, in the browser or through the API.
- From $2.10 in / $10.50 out per 1M tokensSave up to 58%
- OpenAI- and Anthropic-compatible API
- Coding · Reasoning · Agents
What can I help you with?
Use it inClaude CodeClaude desktop app
What is Claude Opus 4.8?
Anthropic describes Claude Opus 4.8 as an Opus model for complex reasoning, agentic coding and professional knowledge work, with an emphasis on planning, code review and sustained task execution.
Use Claude Opus 4.8 in the chat box on this page or call claude-opus-4-8 through the OpenAI-compatible Chat Completions API or Anthropic Messages API. The same UniRoute key can connect your application or an Anthropic-protocol client.
Claude Opus 4.8 continues the Opus line. Anthropic emphasizes stronger coding and debugging, context handling and self-correction for agents that work through a task in stages.
What you can do
What Claude Opus 4.8 is built for.
Coding and code review
Describe a change, paste the relevant code and work through implementation or review comments with follow-up questions.
Analysis and reasoning
Bring documents, requirements or a difficult question into the conversation and ask for a structured explanation of the findings and tradeoffs.
Agent workflows
Use the API to connect the model to tools and build workflows that plan tasks, take actions and review results.
The chat box on this page accepts text only; attachments and tool calls need an API integration.
Specs & variants
The variants you can call, with their options and prices.
| Variant | Model ID | Modes | Inputs | Output | Price |
|---|---|---|---|---|---|
| Claude Opus 4.8 | claude-opus-4-8 | Chat and text generation | Chat messages | Streaming; Thinking; Context window: Not stated on this page | $2.10 input / $10.50 output per 1M tokens |
Pricing
Priced per token. Input and output are billed separately.
| Usage | Official API | UniRoute | Save |
|---|---|---|---|
| Input | $5.00 | $2.10 | Save 58% |
| Output | $25.00 | $10.50 | Save 58% |
| Cache read | $0.50 | $0.21 | Save 58% |
| Cache write (5 min) | $6.25 | $2.625 | Save 58% |
| Cache write (1 hour) | $10.00 | $4.20 | Save 58% |
Official price: platform.claude.com · 2026-11-07
Prices updated 2026-11-08.
Pay as you go from a single USD wallet balance. You see the quote before each request.
API
Call Claude Opus 4.8 through OpenAI- or Anthropic-compatible endpoints.
curl 'https://api.uniroute.app/v1/chat/completions' \ -H 'Authorization: Bearer YOUR_API_KEY' \ -H 'Content-Type: application/json' \ -d '{ "model": "claude-opus-4-8", "messages": [ { "role": "user", "content": "Hello" } ], "stream": false}'curl 'https://api.uniroute.app/v1/messages' \ -H 'Authorization: Bearer YOUR_API_KEY' \ -H 'anthropic-version: 2023-06-01' \ -H 'Content-Type: application/json' \ -d '{ "model": "claude-opus-4-8", "max_tokens": 1024, "messages": [ { "role": "user", "content": "Hello" } ]}'Parameters
| Name | In | Description |
|---|---|---|
model | Body | Model ID to run. |
messages | Body | The conversation as a list of role and content messages. |
max_tokens | Body | Upper limit on generated tokens (Anthropic Messages). |
stream | Body | Set to true to receive the response as a stream. |
anthropic-version | Header | Protocol version header for the Anthropic Messages API. |
Guides
Benchmarks & ranking
Published scores next to the models they were compared with.
| Benchmark | Claude Opus 4.8 | Claude Opus 4.6 | Claude Sonnet 4.6 | GPT-5.4 | Gemini 3.1 Pro |
|---|---|---|---|---|---|
| SWE-bench Verified | 75.2% | 72.5% | 65.3% | 69.8% | 63.1% |
| Terminal Coding (USACO) | 58.0% | 54.2% | 43.6% | 56.5% | 41.2% |
| Agentic Tool Use | 93.5% | 90.1% | 82.4% | 88.7% | 80.5% |
| GPQA Diamond (Reasoning) | 78.4% | 74.9% | 68.2% | 76.1% | 70.3% |
| MMLU Pro | 89.7% | 86.3% | 81.5% | 88.2% | 83.6% |
| Long-Context Retrieval (1M) | 97.8% | 95.2% | 88.1% | 91.5% | 93.4% |
| Code Review Accuracy | 91.2% | 87.6% | 79.4% | 85.3% | 76.8% |
| Max Context Window | 1M+ | 1M | 200K | 128K | 2M |
| Max Output Tokens | 128K | 128K | 64K | 32K | 65K |
Figures published by Anthropic from its own evaluations. Results depend on each benchmark’s configuration.
Use cases
Typical ways to put Claude Opus 4.8 to work.
Plan a repository refactor
Start with the goal and constraints, ask for a plan, then examine the proposed changes before applying them.
Debugging and review
Share symptoms, logs or a diff and ask for likely causes, missing checks and a practical path to a fix.
Document analysis
Turn written material into a summary, compare competing requirements and ask follow-up questions about the reasoning.
A coding assistant in your terminal
Connect Claude Code with a UniRoute key and select claude-opus-4-8. The setup guides below cover compatible clients.
FAQ
Common questions about Claude Opus 4.8.
What is Claude Opus 4.8?
Anthropic describes Claude Opus 4.8 as an Opus model for complex reasoning, agentic coding and professional knowledge work, with an emphasis on planning, code review and sustained task execution. You can use it on this page or through the API.
How do I call Claude Opus 4.8 through the API?
Create a UniRoute API key and send requests with the model ID claude-opus-4-8. The API section shows the OpenAI-compatible Chat Completions and Anthropic Messages request formats.
Can I use it with Claude Code?
Yes. Claude Code uses the Anthropic Messages protocol. Follow the linked setup guide to configure the UniRoute gateway, your key and the model ID.
How much does Claude Opus 4.8 cost?
From $2.10 input / $10.50 output per 1M tokens. Input and output are billed separately based on usage. The Pricing section lists the rates and worked examples (prices updated 2026-11-08).
Does the chat box support attachments or tools?
The chat box accepts text messages. Use an API client for attachments and tool calls; these controls are not part of this page’s chat composer.
Is there a free trial or free credits?
No. Requests are paid from your wallet balance. You can read this page and review the prices without paying, and you need an account with funds to send a message.
Related models
Other versions, models from the same provider and alternatives.
