Claude Opus 5
Last updated: July 25, 2026
Specs
| Vendor | Anthropic |
|---|
| Released | July 24, 2026 |
|---|
| Context window | 1M tokens |
|---|
| Input | $5 / M tokens |
|---|
| Output | $25 / M tokens |
|---|
Benchmarks
Strengths
- #1 in real-world coding: 97.0% SWE-bench Verified (Vals AI) and 43.3% Frontier-Bench v0.1, ahead of Fable 5 and GPT-5.6 Sol
- #1 in agentic knowledge work: GDPval-AA v2 Elo 1861 and Artificial Analysis Intelligence Index 61 at max effort, the highest in the world
- Near or above Fable 5 performance at half the price ($5 / $25 per M tokens), with a 5-level effort selector as a cost/latency lever and a May 2026 knowledge cutoff
Weaknesses
- Behind Mythos 5 on offensive cybersecurity and long-duration autonomous biology, and behind Fable 5 on DeepSWE v1.1 (68.8% vs 69.7%)
- High latency at max effort: 64.5s to first token
- No server-side web fetch and no Priority Tier at launch
Best for
- Production coding agents
- Agentic knowledge work: documents and analysis
- Computer use and desktop automation
- Existing Opus 4.8 workloads: drop-in generational upgrade at the same price
Compare with other AI models