Claude Opus 5: Affordable AI for Coding and Knowledge Work

Anthropic has launched Claude Opus 5, a new AI model that delivers performance close to its flagship Claude Fable 5 at roughly half the cost. Priced at $5 per million input tokens and $25 per million output tokens, Opus 5 is designed for everyday coding, knowledge work, and enterprise workflows. For AI builders evaluating cost-performance trade-offs, this model offers a practical alternative to more expensive frontier models.

What happened

Anthropic introduced Claude Opus 5 as a more affordable alternative to Claude Fable 5, claiming it matches or exceeds Fable 5 on several benchmarks while maintaining the same pricing as its predecessor Opus 4.8. According to Anthropic, Opus 5 outperformed Fable 5 on 10 out of 14 benchmark tests and even surpassed OpenAI's GPT-5.6 Sol in selected evaluations. The model also demonstrated practical capabilities during internal testing, including reconstructing a machine component into a 3D FreeCAD model and building a market data feed for a trading platform in a single session.

Opus 5 is now the default model for Claude Max subscribers and the most capable option for Claude Pro users. Unlike Claude Mythos 5, which focused on advanced cybersecurity, Opus 5 was intentionally not trained to excel in offensive cyber or biological research. It can still help developers detect and fix software vulnerabilities but includes safeguards to block high-risk requests.

Why AI builders should care

The price-performance ratio of Claude Opus 5 is significant for teams building AI-powered products. At $5 per million input tokens and $25 per million output tokens, it costs about half of Fable 5. This makes it viable for high-volume coding tasks, agentic workflows, and knowledge-intensive applications where cost per token directly impacts margins.

Anthropic's positioning comes as US AI companies face competition from lower-cost Chinese models like Kimi K3 and GLM 5.2. Opus 5 offers a middle ground: near-frontier performance without the premium price tag. For startups and enterprises that need reliable AI for daily operations, this could shift purchasing decisions toward value rather than raw capability.

Practical implications

Developers can leverage Opus 5 for coding and knowledge tasks with adjustable reasoning effort, allowing them to balance intelligence, speed, and token consumption. This is useful for applications where latency and cost matter, such as real-time code completion, documentation generation, or data extraction pipelines.

The safety design of Opus 5 also has practical implications. By avoiding training for high-risk domains, Anthropic likely reduced regulatory friction, making the model easier to deploy in enterprise environments with compliance requirements. However, teams working on security-sensitive applications should verify that the safeguards do not block legitimate use cases.

Caveats

All performance and pricing details are based on Anthropic's statements and media reports. Real-world results may vary depending on use case, data, and deployment context. The benchmark comparisons to Fable 5 and GPT-5.6 Sol are vendor-claimed and have not been independently verified. Additionally, while Opus 5 is cheaper than Fable 5, Chinese rivals like Kimi K3 and GLM 5.2 may still offer lower absolute pricing, so teams should evaluate total cost of ownership for their specific workloads.