Opus 5 is my new go-to model

2026-07-25 · Source: Theo - t3․gg · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Advanced, extended

Summary

Anthropic has released Opus 5, a new large language model demonstrating strong performance across various benchmarks, often surpassing Fable 5 in agentic coding and task execution. Priced at \$5 per million input tokens and \$25 per million output tokens, Opus 5 is half the token cost of Fable 5, though its lower token efficiency translates to a real-world cost reduction of 20-25%. Notably, Opus 5 is Anthropic's most aligned and safest model, distilled from Mythos, and supports Zero Data Retention, enabling enterprise adoption previously restricted for Fable. The author's practical experience highlights Opus 5's diligence, precise instruction following, and efficient use of subscription limits, consuming only 12% of a weekly limit for a full day's work, compared to Fable's significantly higher consumption. This positions Opus 5 as a compelling default, bridging the gap between the "wise owl" Fable and the "Rottweiler" 5 6 Soul.

Key takeaway

For AI Engineers or ML teams evaluating new LLMs for daily development and enterprise applications, Opus 5 presents a compelling default. Its balance of Fable's code quality and 5 6 Soul's diligence, combined with Zero Data Retention and significantly slower subscription limit consumption, makes it highly practical. You should integrate Opus 5 into your workflows, especially for tasks requiring precise instruction following and robust code output, and compare its performance against Fable or Soul for specific use cases to validate its fit.

Key insights

Opus 5 balances Fable's quality and Soul's diligence, offering a safer, more cost-effective, and instruction-following model for daily use.

Principles

• Model distillation can enhance safety and alignment.
• Token cost doesn't always reflect real-world task cost.
• Diligence and instruction-following improve model utility.

Method

Anthropic distilled Opus 5 from the larger Mythos model, filtering for desired capabilities and safety, akin to a teacher guiding a student to avoid mistakes and focus on essential knowledge.

In practice

• Use Opus 5 for code merges and daily development.
• Compare models by having them review each other's plans.
• Consider Opus 5 for enterprise with ZDR requirements.

Topics

• Anthropic Opus 5
• Large Language Models
• LLM Benchmarking
• Code Generation
• Model Alignment
• Zero Data Retention
• Cost-Effectiveness

Best for: CTO, VP of Engineering/Data, AI Architect, AI Engineer, Machine Learning Engineer, Director of AI/ML

Related on AIssential

• OPUS 5 CLICK NOW — Claude Opus 5 redefines LLM value, offering superior performance at a significantly lower cost per task.
• Opus 5 vs Fable 5 - Frontier Coding & Agentic Model? | Coding with Claude Code, UI, Backend Game Dev — Fable 5 excels in creative design and speed, while Opus 5 offers cost-effective, thorough iteration for daily agentic coding.
• I hate Opus 5. It’s the best model, anyway. — Opus 5 excels in output quality for design tasks despite a "neurotic" and verbose interaction style.
• Fable 5 Is Twice the Price of Opus 4.8. Here’s When It’s Worth It — Route AI tasks based on complexity and cost-effectiveness, using advanced models only when their superior performance is critical.
• Claude Opus 5 Is THE GREATEST AI Model EVER?! Beats Fable & CHEAPER! (Fully Tested) — Claude Opus 5 offers frontier-class AI capabilities at significantly lower cost than competitors, excelling in complex reasoning and autonomous code generation.
• Anthropic Shipped Four Models in Nine Weeks. Opus 5 Explains the Other Three. — Claude Opus 5 offers near-frontier intelligence at half the price of Fable 5, impacting the value proposition of other models.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Theo - t3․gg.