Technology news and discussion on TalkFuse.
Anthropic debuts Opus 5 at half the price of its Fable sibling
Anthropic has launched Claude Opus 5, saying it approaches the capabilities of its higher-end Fable 5 model while charging half as much per input and output token. The release matters to businesses facing growing AI usage costs, although task-level spending can still differ because models need varying numbers of tokens to complete the same work.
Opus 5 costs $5 per million input tokens and $25 per million output tokens, versus Fable 5’s $10 and $50, and leads Artificial Analysis’s Intelligence Index with a score of 61. Anthropic says it has improved coding, computer-use and long-running knowledge-work performance, while relaxing some source-code vulnerability analysis restrictions but continuing to block penetration testing and exploit generation. The model may produce longer answers unless users specify brevity, and Anthropic is introducing API fallback routing for requests deemed too risky; it also says Opus 5 does not require data retention.
In short
- Opus 5 targets near-frontier performance at half Fable 5’s token price.
- It leads an AI benchmark but may generate longer responses.
- Anthropic eases some safety restrictions while retaining exploit-related safeguards.
Explained simply
Anthropic is an artificial intelligence company that builds software capable of understanding and writing text. Its models are used by businesses for tasks like writing, coding, and analysis. The company has released different versions of its Claude model, each with varying capabilities and costs.
These AI models process text in small units called tokens. Companies are charged based on how many tokens they use, so pricing affects the cost of using AI. When AI models become more capable, companies typically expect to pay more, making cheaper alternatives with similar performance valuable for managing costs.
Anthropic has launched Opus 5, a new version of its Claude model that the company says performs similarly to its more expensive Fable 5 model but costs half as much per token. For organisations using AI regularly, this cost reduction for comparable capability could significantly affect their spending. The improved performance also means organisations may be able to use a single model for more tasks rather than switching between different versions.