Anthropic Launches Claude Opus 5 with Near-Frontier Intelligence at Half the Cost
Anthropic has introduced Claude Opus 5, an AI model that delivers near-frontier intelligence at half the cost of Fable 5. It performs exceptionally well in coding and knowledge work assessments such as Frontier-Bench and GDPval-AA, although it falls short compared to Mythos 5 in cybersecurity. The pricing structure is set at $5 for every million input tokens and $25 for output tokens, making it the primary model on Claude Max and the most powerful on Claude Pro. Opus 5 shows a significant enhancement over Opus 4.8, particularly in scientific research, with notable advancements in organic chemistry and protein-related tasks. It includes improved alignment, reduced deceptive behavior, and protections against misuse. Opus 5 is accessible across all platforms, including Fast mode, and Anthropic has also contributed an extra $20 million to Public First Action.
Key facts
- Claude Opus 5 offers near-frontier intelligence at half the price of Fable 5.
- It sets state-of-the-art results on Frontier-Bench and GDPval-AA.
- On Frontier-Bench v0.1, Opus 5 more than doubles Opus 4.8's performance at lower cost.
- On CursorBench 3.2 at max effort, it performs within 0.5% of Fable 5 at half the cost.
- On ARC-AGI 3, Opus 5 scores three times higher than the next-best model.
- Opus 5 is Anthropic's most aligned model with lowest deceptive behavior rates.
- It remains behind Mythos 5 on cybersecurity and biology research.
- Priced at $5 per million input tokens and $25 per million output tokens.
Entities
Institutions
- Anthropic
- Claude Max
- Claude Pro
- Claude API
- Claude Platform
- Claude Code
- Claude Cowork
- Cursor
- Devin
- Zapier
- JetBrains
- Lovable
- Box
- Public First Action
- Economic Futures Research Fund
- Anthropic Economic Index
Sources
- Anthropic Blog —
- Quartz —