Claude AI 4.8 Update: Better at Math and Coding, But Costs More Tokens

Anthropic just released Claude Opus 4.8, an update to their AI assistant that's causing mixed reactions among users. Think of it like a software update for your phone - some things get better, but others might get worse.

The new version shows clear improvements in math and coding. When tested with complex math problems, it solved them correctly where the previous version failed. For coding, it created a working zombie typing game in one try - something that usually takes multiple attempts with other AI models.

However, there's a major catch: it uses tokens (the currency for AI usage) extremely fast. One user burned through their entire monthly allowance with just a single coding request. For context, tokens are like minutes on a phone plan - when you run out, you have to wait or pay more.

The creative writing abilities haven't improved much. When asked to write stories, the new version produced technically correct but less imaginative content compared to its predecessor. It's like having a writer who follows all the grammar rules perfectly but lacks that creative spark.

Key takeaways: • Better at technical tasks (math, coding) • Much more expensive to use due to high token consumption • No improvement in creative tasks • Price remains at $5 per million input tokens

For users who need AI for technical work and have unlimited budgets, this update is great. But for everyone else, especially creative professionals or those on limited plans, sticking with the older version might be smarter.

This is an AI-generated summary. Read the original article at: https://decrypt.co/370128/claude-opus-4-8-review