Archive · 28 Jul 2026
Anthropic Launches Claude Opus 5
Anthropic has launched Claude Opus 5, a new AI model that offers near-flagship intelligence at half the cost of its top-tier model, Fable 5.
Opus 5 is designed for everyday users, professionals, and developers who need high-value AI capabilities without paying a premium.
It performs almost as well as Fable 5 on coding and computer interaction tasks, with better alignment and fewer refusals.
Opus 5 is available on various subscription tiers, including Claude Pro and Claude Max, offering different levels of access and features.
It is ideal for complex coding, research, document analysis, and automated workflows, while Fable 5 is suited for extreme research and advanced tasks.
Claude Opus 5 is a new AI model launched by Anthropic, offering near-flagship intelligence at half the cost of Fable 5.
Opus 5 performs within 0.5% of Fable 5's peak score on coding benchmarks and surpasses Fable 5 on computer interaction tasks.
Opus 5 achieves a 2.3 score on behavioral audits, with classifiers intervening 85% less often than Fable 5.
Opus 5 features adjustable effort settings from Low to Max Effort, allowing users to dynamically adjust reasoning levels.
Opus 5 is available on various subscription tiers, including Claude Pro and Claude Max, offering different levels of access and features.
- Who
- Anthropic
- What
- Launch of Claude Opus 5 AI model
- When
- July 24
- Where
- Global
- Why
- To provide a cost-efficient, high-performance AI model for everyday users
Claude Opus 5
Claude Fable 5
Cost Efficiency
Claude Opus 5
Opus 5 offers near-flagship intelligence at half the operating cost of Fable 5, making it a high-value model for everyday users.
Claude Fable 5
Fable 5 is positioned as the top-tier model with uncompromised capability, suitable for extreme research and advanced tasks.
Performance
Claude Opus 5
Opus 5 performs within 0.5% of Fable 5's peak score on coding benchmarks and surpasses Fable 5 on computer interaction tasks.
Claude Fable 5
Fable 5 is designed for maximum unconstrained frontier compute, ideal for cutting-edge research and advanced vulnerability exploitation.
Alignment and Refusal Friction
Claude Opus 5
Opus 5 achieves a 2.3 score on behavioral audits, with classifiers intervening 85% less often than Fable 5, reducing frustrating refusals.
Claude Fable 5
Fable 5 may have higher refusal rates due to its more stringent safety measures.
- Token Pricing
- $5.00 per million input tokens, $25.00 per million output tokens
- Speed
- Opus 5 supports Fast Mode at approximately 2.5× default speed
- Subscription Tiers
- Claude Pro, Claude Max 5x, Claude Max 20x
- Coding Benchmark
- Opus 5 performs within 0.5% of Fable 5's peak score
- Computer Interaction
- Opus 5 surpasses Fable 5 on computer interaction tasks
- Alignment Score
- Opus 5 achieves a 2.3 score on behavioral audits
- Dynamic Effort
- Opus 5 features adjustable effort settings from Low to Max Effort
