On September 22, 2026, Anthropic released Claude Opus 5.5 — the first model in its brand-new Claude 5.5 family. The headline numbers are hard to ignore: performance on par with the flagship Claude Fable 5.1 on most tasks, but roughly 40% cheaper to run than its predecessor, Opus 5. Here is what changed, what it costs, and why it matters.
What Is Claude Opus 5.5?
Anthropic describes Opus 5.5 as a major step up from Opus 5, built for the heaviest real-world jobs: agentic coding, computer use, and knowledge work. It is the first release of the 5.5 generation — Sonnet 5.5 and Haiku 5.5 are expected to follow in the coming weeks with many of the same improvements to performance, efficiency, and safety.
The Price Drop, in Plain Numbers
Here is the new API pricing, compared with Opus 5:
- Input tokens: $4 per million (down from $5)
- Output tokens: $20 per million (down from $25)
- Cache reads: $0.20 per million — 60% less than Opus 5. This matters because cache reads make up most of the cost in long agentic coding sessions.
Anthropic estimates typical workloads will cost about 40% less overall, because the model also needs fewer tokens to finish the same task. There is also a “Fast mode” in Claude Code and the Claude Platform that runs up to 2.5x faster, priced at $8 per million input tokens and $40 per million output tokens.
Faster Output, Stronger Benchmarks
At default settings, Opus 5.5 generates output more than 30% faster than Opus 5. On benchmarks shared by Anthropic, it scored 66.4% on Terminal-Bench 4.0 for agentic coding, ahead of Fable 5.1’s 55.8%. On CursorBench it reached 52.5% versus 41.7% for GPT-5.6 Sol — while costing roughly a third as much per task. It also posted an Elo of 1846 on GDPval-AA v2.1 across 44 professional occupations, ahead of GPT-6 Astra’s 1542, along with gains on OSWorld 2.0 for computer use and visual chart recognition.
Real-World Tests: 680,000 Lines of Code in a Day
Beyond benchmarks, Anthropic shared practical examples. An early tester used Opus 5.5 to complete a software migration involving around 680,000 lines of code in less than a day — work the company says would have taken an engineering team weeks. In an internal test, Opus 5.5 and Fable 5.1 both translated the widely used HAProxy networking software from C to Rust and passed nearly all of its regression tests — but Opus 5.5 finished in 9.5 hours compared to 12, at 51% lower cost. Another early tester audited and fixed a 200,000-line codebase in under three hours; Opus 5 reportedly needed more than 20 hours and two and a half times as many tokens.
Better Communication and Stronger Safety
Anthropic says the model communicates more naturally than before: important information comes first, jargon is reduced, and it follows the user’s writing instructions more consistently. On safety, it posted the company’s best scores to date on its automated behavioral audit across nearly 2,000 scenarios. Because its biology and cybersecurity capabilities are comparable to Mythos 5.1, it ships with Fable 5.1-class safeguards — most risky cyber tasks are rerouted to the older Opus 4.8, while vetted organizations can apply for Life Sciences access. External groups including Frontier Design and METR evaluated the model before release.
Where You Can Use It
Claude Opus 5.5 is available now across Anthropic’s platforms: Claude Pro, Max, Team, and Enterprise plans, plus the Claude Platform, Amazon Web Services, Google Cloud, and Microsoft Azure (model identifier: claude-opus-5-5). Anthropic has also raised five-hour usage limits on Pro, Max, and Team plans, and subscribers received a rate-limit reset they can save and use later.
Why It Matters: The AI Price War Heats Up
Opus 5.5 arrives as frontier AI labs compete as fiercely on cost as on capability. Matching a top-tier model’s performance at 40% lower running cost puts direct pressure on rivals’ pricing — and for developers and businesses running AI agents at scale, the savings on long coding sessions are immediate and real.
Conclusion
Claude Opus 5.5 is not just a faster model — it is a cheaper one, and that combination is the real story. With Fable-class performance, output more than 30% faster, sharply lower API prices, and the company’s strongest safety scores yet, Anthropic has set a new bar for what a flagship AI model should cost to run. With Sonnet 5.5 and Haiku 5.5 still to come, the 5.5 family is only getting started.
Explore more stories in our AI Tools & AI News section.
