Skip to content
Ars Technica·

🤖Anthropic Unveils Opus 5: Cheaper Coding AI

Cheaper coding AI that's still good enough?

TL;DR

Anthropic released Opus 5, an update to its popular coding AI model. It offers better performance than OpenAI’s GPT-5.6-Sol and is priced at half of Fable's cost, making it appealing for developers looking to save on AI expenses.

Anthropic just dropped Opus 5, a new coding AI that's cheaper but still competitive with the best out there. If you're a developer who uses AI models like GPT-5.6-Sol or Fable, this is worth checking out. It costs $25 per million output tokens and performs better than its predecessors while being half as expensive as Fable. The catch? While it's great for finding vulnerabilities, it can't match Mythos when it comes to exploiting them. Plus, it doesn’t have all the same data retention policies as Fable. So if you're on a budget but still need top-notch performance, Opus 5 might be your new go-to.

Anthropic Unveils Opus 5: Cheaper Coding AI — Ars Technica

Key Points

1

Opus 5 costs $25 per million output tokens, similar to its predecessor

2

$5 per million input tokens for Opus 5, same as before

3

Performance is comparable or slightly ahead of Fable in coding tasks

4

Avoided cutting-edge training on cybersecurity tasks, lagging behind Mythos

5

Chinese Kimi K3 offers similar performance at $15 per million output tokens

Why It Matters

If you're a developer using GPT-5.6-Sol or Fable for coding tasks and looking to save money without sacrificing too much performance, Opus 5 is your new option. It costs half as much as Fable but still outperforms GPT-5.6-Sol in benchmarks. However, if cybersecurity exploitation is a must-have feature, you might want to stick with Mythos or Fable.

AnthropicOpus 5FableGPT-5.6-Solcost-effective

Frequently Asked Questions

Why does this matter?

If you're a developer using GPT-5.6-Sol or Fable for coding tasks and looking to save money without sacrificing too much performance, Opus 5 is your new option. It costs half as much as Fable but still outperforms GPT-5.6-Sol in benchmarks. However, if cybersecurity exploitation is a must-have feature, you might want to stick with Mythos or Fable.

What happened?

Anthropic released Opus 5, an update to its popular coding AI model. It offers better performance than OpenAI’s GPT-5.6-Sol and is priced at half of Fable's cost, making it appealing for developers looking to save on AI expenses.

Comments

Subscribe to join the conversation...

Be the first to comment

Enjoyed this article?

Get it daily. 7am. Free. Reads in 5 minutes.

Also get