Claude Opus 4.5
Claude Opus 4.5 adds an effort parameter for token-usage control and excels at difficult coding, agentic workflows with computer use and tool use, frontend coding, and real-world web app development, with improvements in general intelligence and vision over previous Opus iterations.
import { streamText } from 'ai'
const result = streamText({ model: 'anthropic/claude-opus-4.5', prompt: 'Why is the sky blue?'})Playground
Try out Claude Opus 4.5 by Anthropic. Usage is billed to your team at API rates. Free users (those who haven't made a payment) get $5 of credits every 30 days.
Providers
Route requests across multiple providers. Copy a provider slug to set your preference. Visit the docs for more info. Using a provider means you agree to their terms, listed under Legal.
| Provider |
|---|
| 200K | 1.5s | 47tps | $5.00/M | $25.00/M | Read:$0.5/M Write: $6.25/M | $10.00/K + input costs | — | 11/24/2024 | ||||
| 200K | 1.3s | 43tps | $5.00/M | $25.00/M | Read:$0.5/M Write: $6.25/M | — | — | 11/24/2024 | ||||
| 200K | 0.7s | 48tps | $5.00/M | $25.00/M | Read:$0.5/M Write: $6.25/M | $10.00/K + input costs | — | 11/24/2024 |
P50 throughput on live AI Gateway traffic, in tokens per second (TPS). Visit the docs for more info.
P50 time to first token (TTFT) on live AI Gateway traffic, in milliseconds. View the docs for more info.
Direct request success rate on AI Gateway and per-provider. Visit the docs for more info.
More models by Anthropic
| Model |
|---|
| 1M | 0.8s | 97tps | $5.00/M | $25.00/M | Read:$0.5/M Write: $6.25/M | $10/K + input costs | — |    | 04/16/2026 | ||||
| 1M | 0.6s | 52tps | $3.00/M | $15.00/M | Read:$0.3/M Write: $3.75/M | $10/K + input costs | — | | 02/17/2026 | ||||
| 1M | 0.8s | 52tps | $5.00/M | $25.00/M | Read:$0.5/M Write: $6.25/M | $10/K + input costs | — | | 02/05/2026 | ||||
| 200K | 0.4s | 115tps | $1.00/M | $5.00/M | Read:$0.1/M Write: $1.25/M | $10.00/K + input costs | — | | 10/15/2025 | ||||
| 1M | 0.7s | 55tps | $3.00/M | $15.00/M | Read: $0.3/M Write: $3.75/M | $10.00/K + input costs | — | | 09/29/2025 | ||||
| 1M | 0.7s | 75tps | $3.00/M | $15.00/M | Read: $0.3/M Write: $3.75/M | $10.00/K + input costs | — | | 05/22/2025 |
About Claude Opus 4.5
Claude Opus 4.5 became available on Vercel AI Gateway on November 24, 2024. The model adds a new effort parameter that affects all token types and governs how much token usage the model applies per request. It defaults to high and operates independently of the thinking budget, giving you a separate lever for token efficiency without adjusting extended thinking configuration.
Difficult coding tasks and agentic workflows are core strengths, especially those combining computer use and tool use. The model handles context from external memory files effectively. Frontend coding and design for real-world web applications are established strengths of this checkpoint.
To use the model, set anthropic/claude-opus-4.5 and configure effort through providerOptions in the AI SDK. The effort setting is independent of the thinking budget. Set both separately to manage the cost-quality tradeoff for different request types in the same application.
Anthropic designed Opus 4.5 for the demanding end of the agent capability spectrum: workflows involving computer use, external tool calls, memory file management, and frontend development where design quality and code correctness both matter.
What To Consider When Choosing a Provider
- Configuration: The effort parameter (high by default) affects all token types for a request. Setting it in
providerOptions.anthropic.effortlets you manage token usage per-request when running mixed workloads with varying complexity. - Zero Data Retention: AI Gateway supports Zero Data Retention for this model via direct gateway requests (BYOK is not included). To configure this, check the documentation.
- Authentication: AI Gateway authenticates requests using an API key or OIDC token. You do not need to manage provider credentials directly.
When to Use Claude Opus 4.5
Best For
- Agentic workflows combining computer use and tool use: Highlighted as a core strength for tasks requiring both capabilities in combination
- Frontend coding and design for real-world web apps: Production-quality output on complex UI development tasks
- Demanding reasoning and complex problem solving: The
effortparameter (high by default) lets you apply maximum token usage per request for the hardest analytical work - External memory file handling: The model excels at managing context across long agentic sessions with memory file access
- Tasks where effort-level control matters: Workloads mixing simple and complex requests that benefit from per-request token usage management
Consider Alternatives When
- Fastest interactive response: Haiku 4.5 or Sonnet 4 variants are better suited for low-latency use
- Sonnet-level depth suffices: A Sonnet model handles tasks that don't require Opus-level reasoning at lower cost
- 1M token context window: Context size came to Opus models with version 4.6
Conclusion
Claude Opus 4.5 is built for the demanding intersection of agentic workflows: computer use, external tool calls, memory management, and frontend development, with the effort parameter giving teams explicit cost control on top of the model's capabilities. For organizations pushing on the complexity ceiling of what agents can handle autonomously, it's the right Opus checkpoint prior to 4.6.
Frequently Asked Questions
What is the effort parameter in Claude Opus 4.5 and how does it work?
The effort parameter controls how much token usage the model applies when responding to a request. It defaults to high, affects all token types, and is set independently of the thinking budget via
providerOptions.anthropic.effortin the AI SDK.How do I configure the effort parameter through AI Gateway?
Under
providerOptions.anthropicin your AI SDK v5 call, pass aneffortlevel oflow,medium, orhighalongside the model stringanthropic/claude-opus-4.5.What agentic capabilities does Opus 4.5 excel at specifically?
Difficult coding tasks, agentic workflows combining computer use and tool use, and complex multi-step problem solving. The model handles tasks requiring sustained reasoning across long contexts and multi-turn interactions.
How does effort differ from the thinking budget?
They are independent parameters. The effort parameter controls overall token usage across all token types; the thinking budget specifically controls extended reasoning tokens. You can set both separately to tune the cost-quality tradeoff.
What vision improvements does Opus 4.5 include?
Opus 4.5 includes improvements in general intelligence and visual reasoning, handling image analysis, diagram interpretation, and document understanding with stronger accuracy than previous Opus versions.
Does Opus 4.5 support the 1M token context window?
No. The 1M token context window came to Opus models with Claude Opus 4.6.
Is Claude Opus 4.5 suitable for frontend web development work?
Yes, frontend coding and design for developing real-world web applications is specifically listed as an established strength of this model.