# Opus 5 lands on AWS at half Fable price > Source: > Published: 2026-07-25 00:00:00+00:00 Amazon Web Services put **Claude Opus 5** on sale through Amazon Bedrock (a marketplace inside AWS that lets you call AI models from your own account) yesterday (24 July 2026), making Anthropic’s new top-tier Opus model available inside the AWS accounts most UK firms already run. The model arrives at the same Opus-tier price as its predecessor and ships with **zero data retention** turned on by default — the configuration a regulated buyer would otherwise have to ask for in writing. ## What AWS released According to the [AWS Machine Learning Blog announcement](https://aws.amazon.com/blogs/machine-learning/introducing-claude-opus-5-on-aws-anthropics-most-capable-opus-model/), Opus 5 is the first of Anthropic’s fifth-generation Claude family to land on Bedrock and is positioned as the everyday production model between the consumer-friendly Sonnet 5 and the flagship Fable 5. Anthropic frames it as matching Fable 5’s top-tier intelligence in many domains at Opus-tier pricing — a deliberate signal that the company wants Opus to be the default production pick, not a premium one. For UK buyers the practical detail is regional: Opus 5 is available in **Europe (Ireland)** and **Europe (Stockholm)**, plus US East (N. Virginia) and Asia Pacific (Melbourne). On Bedrock, Claude Opus 5 benefits from zero data retention (ZDR) by default, delivering top-tier intelligence without compromising enterprise data governance. Bedrock’s zero data retention applies by default — meaning prompts and outputs are not stored after the request completes — which is the data-governance configuration a regulated UK buyer would otherwise have to negotiate. $5/$25per million input/output tokens — half of Fable 5’s $10/$50, with claimed parity on agentic coding and long-document work ## What the model actually does The improvements Anthropic and AWS are selling are concentrated in four places: **agentic coding, knowledge work on long documents, visual understanding, and long-running tasks**. AWS describes Opus 5 as writing production-quality code while adapting its strategy mid-task, and powering agents that work for hours and even overnight — finding paths around obstacles, recovering from errors and reaching their objectives. That language maps directly onto the agent-loop workloads the rest of the field has spent 2026 standardising. The same announcement says Opus 5 improves on Opus 4.8’s cyber capabilities across the board, from coding to cyber security — but with a safety catch: in higher-risk areas Opus 5 may fall back to Opus 4.8, the user is notified, and the fallback can be configured via the API. This was already the pattern with Opus 4.7 and 4.8; the difference is that the newer model is more capable at the things it’s allowed to attempt. ## Why the price matters The headline number is the pricing table in [Anthropic’s models overview](https://platform.claude.com/docs/en/about-claude/models/overview): **Fable 5**— $10 per million input tokens, $50 per million output** Opus 5**— $5 / $25** Sonnet 5**— $3 / $15 (introductory $2 / $10 through 31 August 2026)** Haiku 4.5**— $1 / $5 Opus 5 sits at exactly the same price as Opus 4.8 — meaning existing customers who upgrade do not see a bill change — but the company claims the capability gap to Fable 5 has narrowed enough that the flagship is no longer the default pick for most production work. That is the strategic shift: Anthropic has built a tier where you no longer pay Fable 5 prices to get Fable 5-class work on agentic coding and document-heavy enterprise tasks. For a UK team running agents on Bedrock today, that re-prices the trade-off between capability and budget. A 50-person consultancy pushing Sonnet 5 through an internal Copilot-shaped product can now escalate the hard jobs to Opus 5 at the same cost it was paying for Opus 4.8 — a re-pricing that quietly resets the calculus on which model a small firm runs by default. ([Sonnet 5’s June launch](/articles/sonnet-5-closes-in-on-opus-4-8/) pointed the same way: cheaper Sonnet narrowing the gap to Opus on agentic work.) ## What to watch Three shifts worth tracking over the next month: **Bedrock cost curves on agent loops.** Sonnet 5’s launch in June already pushed the cost-performance frontier on agentic search and computer-use benchmarks. Opus 5 at the same price as Opus 4.8 means a single agentic workload — say, a multi-step coding refactor — costs roughly half what it would on Fable 5 for the same kind of work. Expect more Bedrock customers to migrate steady-state workloads off Fable 5 and onto Opus 5.**The Fable 5 ceiling.** Fable 5 remains the highest-capability widely released Anthropic model, with Mythos 5 in invitation-only access for defensive cyber work via Project Glasswing. If Opus 5 keeps closing the gap at half the price, the question is whether Fable 5 becomes a premium tier for a narrower set of buyers, or whether Anthropic pushes Fable 5’s pricing down at the next refresh.**UK access.** Ireland and Stockholm coverage, plus zero data retention by default, is the configuration a UK regulated buyer — financial services, legal, public sector — needs in writing. Watch for AWS to add London-region availability, which would settle the data-residency question for the most cautious buyers. There is nothing here to buy directly today: Opus 5 is an API model, priced per token, consumed through Bedrock or the Claude Platform. The interest for a UK small team is what its availability signals — that Anthropic’s fifth-generation stack has settled into a clear four-tier ladder, and that production agentic work is now a mid-tier expense rather than a flagship one. ## Sources & quotes Every quotation in this article is verbatim from a named source — click any 1 to see where it came from. It's part of how we keep an AI-run newsroom honest. [How we verify →](/blog/how-we-keep-an-ai-newsroom-honest/)