Anthropic launched Claude Opus 5 on July 24, 2026 — recording the highest-ever score (61) on the AGI Artificial Intelligence Index and 79.2% on SWE-bench, cementing its position as the strongest general-purpose reasoning model available.
Benchmarks
AAII 61 is the highest score ever recorded on Anthropic’s flagship index. SWE-bench 79.2% places it at the frontier of software engineering capability. Early independent tests confirm the scores translate into real productivity gains on complex codebases.
Pricing
Opus 5 launched at $5/M input and $25/M output — a 50% price cut versus its predecessor, signaling Anthropic’s willingness to price aggressively at the top end as competition intensifies.
Positioning
With the June Fable 5 suspension behind it, Opus 5 is Anthropic’s flagship response to GPT-5.6 Sol and Gemini 3.5 Pro. The release is complemented by the broader Claude 5 family — Opus 5, Fable 5, Sonnet 5, Haiku 5 — covering every segment of the market.
Should You Upgrade?
For complex software engineering, deep reasoning, and high-stakes enterprise analysis: yes. For routine tasks, Sonnet 5 or Haiku 5 deliver comparable value at lower cost. The tiering logic remains the smart buying strategy.
Related: AI Model Comparison 2026 · AI Models in 2026: Complete Guide
