# Anthropic launches Opus 5 at half the price of Fable New AI model maintains performance while reducing costs - **Source:** [www.theregister.com - Articles](https://brifing.si/source/2c34ea50-d341-4a10-925d-12c012a9ad10) - **Published:** 2026-07-24 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/858b2dc1-ea11-4e87-8da4-95e4c11f30ea ## Affordable high performance Anthropic's new Opus 5 model offers similar intelligence to Fable 5 at half the price. Input tokens priced at $5, output tokens at $25. This provides significant savings for businesses struggling with rising AI costs, creating a more cost-effective alternative compared to competing models. ## Cost advantage per task According to Artificial Analysis data, Opus 5 is 26% more cost-effective than Fable for Intelligence Index tasks. With an average cost of $2.03 per task, it ranks as the third most economical model after GPT-5.6 Sol and Kimi K3, standing out with its performance-cost balance. ## System prompt optimization Anthropic simplified Opus 5's system prompt by 80%, allowing users to add more context. Engineer Thariq Shihipar notes that the claude doctor command provides automatic optimization. However, the model's responses may be more verbose than previous versions. ## Security and ethical improvements Opus 5 achieves 80% success rate in detecting vulnerabilities in open source code but remains weak in exploiting them. Cybersecurity classifiers are tuned to be less restrictive, blocking malicious activities while permitting beneficial uses, striking a balance between security and functionality. ## Reduced refusal rates The model refuses fewer requests in biology and cybersecurity tasks. Through an automated fallback function in the API, requests deemed too dangerous are routed to simpler models rather than being completely blocked. Anthropic describes Opus 5 as its 'most aligned model to date.' ## Attractive features for businesses One of Opus 5's key business advantages is the absence of data retention requirements. It offers major improvements in long-horizon knowledge work, agentic coding, and computer use, combined with performance boosts, creating an attractive package for enterprise users. ## Sources covering this story - https://www.theregister.com/ai-and-ml/2026/07/25/anthropic-debuts-opus-5-at-half-the-price-of-its-fable-sibling/5278630 — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._