started · updated
Anthropic launches Claude Opus 5 AI model at half the price of Fable 5
Anthropic released its new Claude Opus 5 large‑language model on 24 July 2026. The model is priced at $5 per million input tokens and $25 per million output tokens – exactly half the cost of the flagship Claude Fable 5 – and is offered through the company’s API (claude‑opus‑5), AWS Bedrock, and the Claude Platform.
Anthropic says Opus 5 delivers notably higher performance on several internal benchmarks. It achieved 43.3 % on Frontier‑Bench v0.1, more than double the 21.1 % score of its predecessor Opus 4.8, and recorded 30.2 % on the ARC‑AGI‑3 test, roughly three times the next‑best model. The model also scores within 0.5 % of Fable 5 on CursorBench 3.2 when run at maximum effort, while using roughly half the per‑task cost.
The model supports a context window of up to one million tokens and includes an adjustable “effort” setting with five levels, allowing developers to trade off reasoning depth against token consumption. Anthropic highlights enhanced capabilities for coding, agentic workflows, and knowledge‑work tasks, and notes that the AWS deployment offers zero‑data‑retention options and reduced safety‑classifier triggers for higher‑risk cybersecurity queries.
Entities
AWS (Amazon Web Services) · Amazon Bedrock · Anthropic · Claude Fable 5 · Claude Max · Claude Opus 5 · Dianne Penn · Frontier‑Bench · OpenAI
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 13 SOURCES] Claude Opus 5 performs comparably to Claude Fable 5 on key benchmarks. www.infomagazine.ma · www.world-today-journal.com · dxmagazine.jp · fr.businessam.be · genk.vn · +8 more
- [● 11 SOURCES] Anthropic’s pricing strategy positions it against OpenAI’s consumer‑scale model in a competitive AI market. www.infomagazine.ma · www.world-today-journal.com · dxmagazine.jp · fr.businessam.be · genk.vn · +6 more
- [● 11 SOURCES] Dianne Penn said Anthropic aims to make AI accessible across generations and provide value‑focused pricing for enterprises. www.infomagazine.ma · www.world-today-journal.com · dxmagazine.jp · fr.businessam.be · genk.vn · +6 more
- [● 14 SOURCES] Claude Opus 5 triggers safety classifiers about 85 % less often than Claude Fable 5. www.infomagazine.ma · www.world-today-journal.com · dxmagazine.jp · fr.businessam.be · genk.vn · +9 more
- [● 14 SOURCES] Anthropic launched Claude Opus 5 on 24 July 2026. www.infomagazine.ma · www.world-today-journal.com · dxmagazine.jp · fr.businessam.be · genk.vn · +9 more
- [● 14 SOURCES] Claude Opus 5 is priced at $5 per million input tokens and $25 per million output tokens, roughly half the cost of Claude Fable 5. www.infomagazine.ma · www.world-today-journal.com · dxmagazine.jp · fr.businessam.be · genk.vn · +9 more
- [● 11 SOURCES] Claude Opus 5 is the default model for Claude Max and the most capable option in Claude Pro. www.infomagazine.ma · www.world-today-journal.com · dxmagazine.jp · fr.businessam.be · genk.vn · +6 more
- [● 13 SOURCES] Claude Opus 5 is not trained for offensive cyber‑operations and is less capable than Mythos 5 in that domain. www.infomagazine.ma · www.world-today-journal.com · dxmagazine.jp · fr.businessam.be · genk.vn · +8 more
- [● 3 SOURCES] Dianne Penn is Anthropic’s Head of Product Management. www.infomagazine.ma · www.world-today-journal.com · genk.vn
- [● 3 SOURCES] Claude Opus 5 includes an 'effort' setting with five levels that let users trade off intelligence versus token consumption. www.connect.ro · dxmagazine.jp · webrazzi.com
- [● 2 SOURCES] Claude Opus 5 supports a context window of up to 1 million tokens. www.connect.ro · dxmagazine.jp
- [● 3 SOURCES] Claude Opus 5 includes an adjustable ‘effort’ setting with five levels to trade off reasoning depth against token consumption. www.connect.ro · www.it-boltwise.de · dxmagazine.jp