Skip to content
Ars Technica·

🤖OpenAI and Anthropic Release Cost-Effective AI Models

New models aim to slash costs for enterprise users

TL;DR

OpenAI and Anthropic have released new AI models aimed at reducing costs for enterprise users. Opus 5.5 offers 20% fewer tokens and 30% faster output than its predecessor, while GPT-6 Sol and Luna cut costs by half.

OpenAI and Anthropic have released new AI models, Opus 5.5 and GPT-6 Sol and Luna, respectively, focusing on cost efficiency rather than groundbreaking capabilities. Opus 5.5, the latest version from Anthropic, offers a 20% reduction in input and output tokens and a 30% speed boost compared to its predecessor. GPT-6 Sol and Luna, meanwhile, are designed to be half as expensive to use as their predecessor, GPT-6 Astra. These updates are crucial for developers and enterprises looking to optimize their AI usage and reduce costs. The real story is in the numbers: Opus 5.5 offers significant savings for typical workloads, while GPT-6 Luna's API pricing is $0.10 and $0.50 per 1 million input and output tokens, respectively.

OpenAI and Anthropic Release Cost-Effective AI Models — Ars Technica

Key Points

1

Opus 5.5, the latest from Anthropic, uses 20% fewer tokens and generates output 30% faster than its predecessor.

2

GPT-6 Sol and Luna are half as expensive to use as GPT-6 Astra, with API pricing at $2 and $10 per 1 million input and output tokens.

3

GPT-6 Luna's API pricing is $0.10 and $0.50 per 1 million input and output tokens, respectively, making it the most cost-effective option.

4

Opus 5.5 performs better in coding and knowledge work compared to GPT-6 Astra in some benchmarks.

5

GPT-6 Sol and Luna were trained with similar methods to GPT-6 Astra, focusing on cost and efficiency improvements.

Why It Matters

If you're an enterprise user looking to optimize AI costs, these new models from OpenAI and Anthropic are a game changer. Opus 5.5 offers significant savings with 20% fewer tokens and 30% faster output, while GPT-6 Luna's $0.10 and $0.50 API pricing makes it the most cost-effective option. However, the savings only pencil out for larger workloads, so smaller databases should stay put.

OpenAIAnthropicAI modelscost-effectiveenterprise

Comments

Subscribe to join the conversation...

Be the first to comment

Enjoyed this article?

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

Join 3,488 builders reading daily.

Also get