Frontier Models
Anthropic Releases Claude Opus 5 as No. 2 Model Behind Fable 5
The model provides step-change gains in agentic tasks at the same pricing as Opus 4.8 and half the cost of the flagship Fable 5 with immediate availability on major platforms.
Claude Opus 5 is Anthropic's new model that ranks as the company’s No. 2 model behind Fable 5 with gains in agentic work at the same pricing as Opus 4.8.
Anthropic has released Claude Opus 5 at the same pricing level as the previous Opus 4.8 model. The new model is designed to provide substantial improvements in performance for specific types of tasks. These tasks include agentic coding, computer use and knowledge work. The release allows users to access advanced capabilities without changing their budget for model usage. The decision to keep pricing the same demonstrates a commitment to accessibility for advanced AI tools. Users who were using the previous version can transition to the new one seamlessly in terms of cost.
The positioning of the model as the second in the lineup offers a middle ground for users. Users can choose between the top model and this one depending on their needs. This approach provides flexibility in the model selection process for different workloads. Organizations can now match model capability more closely to the demands of their specific projects.
What background leads to the introduction of Claude Opus 5?
The Opus series has been a key part of Anthropic's offerings for high capability models. The previous version Opus 4.8 set a certain standard for pricing and performance. The new version builds on that foundation with enhanced capabilities in key areas. The company has maintained the pricing structure to ensure continuity for existing users. This decision reflects a strategy to make advanced models more accessible while advancing the technology.
The series has evolved to address increasing demands for models that can handle extended sequences of operations. Previous iterations established benchmarks in reasoning that the latest release aims to surpass in targeted domains. The background includes a focus on delivering practical improvements that translate directly to real world applications.
What new features and improvements are included in Claude Opus 5?
Claude Opus 5 delivers a step change improvement over the prior version in the series. The largest gains are seen in deep reasoning tasks. Additional gains are in agentic and long horizon tasks. Test time compute scaling also sees benefits from the update. These improvements make the model particularly suited for complex agentic coding and enterprise work. The model is recommended for such workloads according to the company documentation.
The enhancements focus on areas where sustained performance over multiple steps is required. This allows the model to maintain coherence and accuracy across longer interactions. The update represents a deliberate effort to advance capabilities in domains that benefit from iterative processing and decision making.
Claude Opus 5 is a step-change improvement over Claude Opus 4.8, with the largest gains in deep reasoning, agentic and long-horizon tasks, and test-time compute scaling.Anthropic
How does the pricing of Claude Opus 5 compare to other models?
The pricing for Claude Opus 5 remains at five dollars per million input tokens. The output pricing is twenty five dollars per million output tokens. This matches the pricing for Claude Opus 4.8 exactly. The cost is half that of the flagship Claude Fable 5 model. This allows for frontier level intelligence at a reduced expense for users.
Maintaining the same pricing structure supports predictable budgeting for teams that rely on consistent model costs. The comparison to the flagship model highlights an option for users who require high performance without the premium associated with the top tier. This tiered approach enables more precise allocation of resources based on task complexity.
What are the technical specifications of Claude Opus 5?
The model comes with a context window of one million tokens. This serves as both the default and the maximum size. The maximum output tokens are set at one hundred twenty eight thousand. Thinking is enabled by default in the model configuration. These specifications support the long horizon tasks that the model is optimized for.
The large context window allows processing of extensive inputs in a single session. This reduces the need for chunking or multiple calls in scenarios involving large datasets or code repositories. The default thinking mode contributes to more deliberate output generation on complex queries.
| Model | Input Price | Output Price | Context Window | Key Use Case |
|---|---|---|---|---|
| Claude Opus 5 | $5/MTok | $25/MTok | 1M tokens | Agentic coding and enterprise work |
| Claude Opus 4.8 | $5/MTok | $25/MTok | Not specified | Previous version |
| Claude Fable 5 | Twice that of Opus 5 | Twice that of Opus 5 | Not specified | Highest capability workloads |
On which platforms is Claude Opus 5 available?
The model is available immediately on several platforms. These include the Claude API and Amazon Bedrock. Additional platforms are the Claude Platform on AWS and Google Cloud. Microsoft Foundry also supports the model right away. This broad availability ensures that users across different environments can access the model without waiting for updates. The immediate rollout supports timely adoption in various enterprise settings.
- Claude API
- Amazon Bedrock
- Claude Platform on AWS
- Google Cloud
- Microsoft Foundry
The distribution across cloud providers reduces dependency on a single infrastructure. Teams can select the platform that aligns with their current architecture and compliance requirements. Immediate access means development and testing can begin without delay following the announcement.
What are the market implications of this release?
The release provides an option for organizations that need high performance but at a lower cost than the top model. This could influence how companies allocate their AI budgets. Stakeholders in the AI space may see this as a way to scale agentic applications more affordably. The model supports workloads that require deep reasoning without the full flagship expense.
Market participants may adjust their procurement strategies to incorporate this tier for intermediate needs. The pricing parity with the prior version minimizes disruption for ongoing projects. This release could accelerate adoption of agentic systems by lowering the entry cost for capable models.
What can be expected next from Anthropic in this area?
Further developments may build on the test time compute scaling features. The company continues to advance its models in the frontier category. The strategy of offering tiered models allows for a range of options for different user needs. This approach is likely to continue with future releases.
Continued emphasis on agentic capabilities suggests ongoing investment in models that can operate autonomously over extended periods. The pattern of maintaining pricing while improving performance may persist as a competitive tactic. Users can anticipate additional refinements in context handling and output quality in subsequent updates.
Frequently asked
What is the pricing for Claude Opus 5?
Claude Opus 5 costs $5 per million input tokens and $25 per million output tokens. This pricing is the same as for Claude Opus 4.8.
Sources
- Anthropic — Claude Opus 5 is a step-change improvement over Claude Opus 4.8, with the largest gains in deep reasoning, agentic and long-horizon tasks, and test-time compute scaling. Pricing is unchanged from Opus 4.8.
- Anthropic — Claude Opus 5 pricing is $5 / MTok input and $25 / MTok output, same as Claude Opus 4.8.
- Anthropic — If you're unsure which model to use, start with Claude Opus 5 for complex agentic coding and enterprise work. For workloads that need the highest available capability, use Claude Fable 5.