Anthropic Unveils Cheaper Claude Opus 5
Synopsis
The new AI model outperforms rivals on coding and knowledge tasks while targeting cost-conscious enterprise customers
Key Highlights
- Anthropic launched Claude Opus 5, calling it the best-performing and lowest-cost AI model it’s created.
- Opus 5 beats Claude Fable 5 on all coding and knowledge work benchmarks at half the cost.
- The rollout comes as companies seek lower-cost AI models that deliver a more obvious ROI.
Anthropic released Claude Opus 5 on Friday, calling it the best-performing and most cost-effective model from the company across multiple industry benchmarks.
The firm says that Opus 5 beats Claude Fable 5 for coding and knowledge work benchmarks, and is developed to be used in everyday life.
Lower-Cost AI for Businesses
Opus 5 costs US$5 per million input tokens and US$25 per million for output tokens, which is half of the pricing for Fable 5. Anthropic stated that this launch comes as companies are more budget-conscious and reluctant to bet on pricey AI models without any return on investment clearly spelled out.
The company is also under pressure due to lower-cost AI alternatives from Microsoft, Amazon, Google and the board of other Chinese startups.
Customers are searching for value, and that lower-price models are relevant only if they deliver identical quality to larger models, Anthropic Head of Product Management for Research Dianne Penn said.
Cybersecurity Remains Top Of Mind With Mythos 5
Anthropic said Opus 5 is not their most capable model with respect to these potentially dangerous dual-use abilities, including cybersecurity.
Amid that backdrop, the company has been under intense scrutiny since its release in April of Claude Mythos Preview, claiming the model was highly effective at finding and exploiting software vulnerabilities.
Anthropic launched Mythos 5 and Fable 5 in June, but Fable 5 is meant for general use. The company disabled access to both models later due to an export control notice issued by government authorities regarding national security. Export controls were lifted after about two weeks, enabling Anthropic to restore access to both models.
Penn said Anthropic can work with government agencies for pre-deployment testing and that Mythos 5 is still superior when it comes to offensive cybersecurity experiments as well as biology research compared with Opus 5.
Source: Reuters
At Inspirepreneurs Magazine, covering entrepreneurship, business failures, and the human stories behind the world's most ambitious founders. She writes at the intersection of strategy and storytelling.
