AT A GLANCE
Faster output, lower developer rates and stronger coding scores are welcome. They do not make the biggest model the right choice for every task. Start with the business process, test the result and total cost..... and control what the AI can read and change.

Anthropic’s Fable 5.2..... faster, better at coding and cheaper!
Anthropic reports up to 30% faster output and 20% lower developer usage rates than Fable 5.1. Anthropic’s announcement (opens a new tab)
Its FrontierCode 2.0 coding score jumps from 32.4% to 49.1%. OpenAI is close behind on that same test..... GPT-6 Astra scores 42.8%.
Artificial General Intelligence (AGI) looks a little closer. Ethan Mollick calls this “almost” AGI (opens a new tab) under his definition of matching human experts across most computer-based intellectual tasks.
His experiments show impressive coding, research and creative work..... but still need human judgement to recognise when results fall short.
BUT..... you do NOT need AGI for every task.
Processing an order against agreed prices, stock availability and credit limits does not require the world’s cleverest model.
Conventional automation can enforce those rules, while a suitable AI model interprets the customer’s email and refers exceptions for review.
Even Anthropic recommends starting with Opus 5.5 for most coding and knowledge tasks, moving to Fable when the work demands it. Anthropic’s guidance (opens a new tab)
Start with YOUR business requirement. Test accuracy, reliability, speed and total cost. Control what the AI can read and change.
Fit for purpose..... with security built in!
THE SOURCE / FURTHER READING
Want the full feature?
You’ve read the TOP STORIES interpretation. Continue to the publisher for the source feature or detailed guidance.
Read the original feature at Computerworld (opens a new tab)