[AINews] Claude Opus 5: Fable-level performance at Opus price (half Fable)
Quick Answer
Anthropic's Claude Opus 5 has launched, outperforming Claude Fable 5 in benchmarks with an ECI of 159, while reducing task costs by 20%.
Quick Take
Despite claims of being 'close' to Fable, independent evaluations confirm its superior coding performance and efficiency, igniting discussions on model evaluation standards.
Key Points
- Claude Opus 5 achieves an ECI of 159, slightly below Fable 5's 161.
- Opus 5 reduces cost per task by 20%, enhancing efficiency.
- Independent evaluations praise Opus 5's coding capabilities over Fable.
- User feedback highlights Opus 5's browser automation skills.
- Debate continues on the adequacy of current benchmarking methods.
DeepSignal Analysis
What happened
Anthropic released Claude Opus 5, which reportedly outperforms Claude Fable 5 in benchmarks with an ECI of 159, while also reducing task costs by 20%. Independent evaluations corroborate its superior coding performance, although some users argue that the ECI score does not fully reflect its practical improvements. The launch has sparked discussions about the validity of current model evaluation standards.
Key evidence
- Claude Opus 5 achieved an ECI of 159, slightly below Fable 5's ECI of 161, according to Epoch AI Research.
- Independent evaluations indicate that Opus 5 outperformed Fable 5 by nearly 150 Elo on the AA-Briefcase benchmark while reducing cost per task by 20%.
- User feedback suggests that Opus 5's practical improvements are not adequately captured by its ECI score, with some calling for more rigorous public benchmarks.
Why it matters
The launch of Claude Opus 5 is significant as it highlights ongoing competition in AI model performance, particularly in coding tasks. The discrepancies between benchmark scores and user experiences raise questions about the reliability of current evaluation methods. As AI applications increasingly demand practical efficiency, understanding these models' real-world capabilities becomes crucial for developers and businesses alike.
Source Excerpt
ain't nobody beats Anthropic at distilling Fable!
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from Latent Space
See more →![[AINews] SpaceXAI launches Grok 4.5, first Opus-class model post Cursor acquisition](https://substackcdn.com/image/fetch/$s_!8D6O!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fpbs.substack.com%2Fmedia%2FHMuQw2BXUAAJaQd.png)
[AINews] SpaceXAI launches Grok 4.5, first Opus-class model post Cursor acquisition
SpaceXAI has launched Grok 4.5, a new coding-focused model that is 3x larger than Grok 4.3, priced at $2 per million input tokens. Positioned as an Opus-class model, it aims for efficiency and speed, outperforming competitors like GPT-5.6 and Opus 4.8 in cost-effectiveness.
