NewsBite analysis · AI, Programming
MarkTechPost
Frontier-Class Agentic Coding and Computer Use at Unchanged Opus Pricing
Anthropic has officially launched Claude Opus 5 as the successor to Claude Opus 4.8, maintaining the same pricing structure of $5 per million input tokens and $25 per million output tokens. Opus 5 is now positioned to rival Claude Fable 5 at a lower cost while integrating several key changes, such as default thinking capabilities, which previously required specific prompts to activate. A notable aspect is the introduction of breaking changes affecting developer requests, particularly around the handling of thinking settings and the need to adjust existing parameters for optimal performance. Furthermore, Opus 5 has quickly demonstrated its capability through rigorous benchmarks, illustrating marked improvements across various task evaluations.
MonitorConfidence: MediumShort term horizon
Published 24 Jul 2026, 22:40 UTC5 min read
Visit original article ↗Lead image — MarkTechPost
Summary
TL;DR
- Launch of Claude Opus 5 with unchanged pricing of $5/$25
- Default thinking feature now enabled on all requests
- Breaking changes affect developers requiring prompt adjustments
- Significant performance improvements over previous models
- Opus 5 achieves high scores in major benchmark tests
Interesting facts
- Opus 5 achieved a score of 70.57% on OSWorld 2.0, a substantial increase from Opus 4.8's performance.
- On the Gray Swan benchmark, attacker success rates improved significantly compared to earlier models.
- The pricing model remains unchanged despite enhanced capabilities and benchmarks.