cd /news/large-language-models/openai-s-new-gpt-6-sol-and-luna-mode… · home topics large-language-models article
[ARTICLE · art-137555] src=macrumors.com ↗ pub= topic=large-language-models verified=true sentiment=↑ positive

OpenAI's New GPT-6 Sol and Luna Models Bring Astra Improvements to Cheaper Tiers

OpenAI launched GPT-6 Sol and GPT-6 Luna, two models following GPT-6 Astra, with GPT-6 Sol priced at $2 per million input tokens and $10 per million output tokens — half the previous rate — and GPT-6 Luna at $0.10 per million input tokens and $0.50 per million output tokens, down from $0.20 and $1.20. OpenAI says GPT-6 Sol makes "about half as many mistakes" as GPT-5.6 Sol and matches or beats Claude Fable 5.1 on some benchmarks, while Anthropic released Opus 5.5 the same day, which outperforms Astra on some coding and knowledge work benchmarks. Both models roll out today in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, with GPT-6 Luna also available to Free and Go users in the desktop app, and OpenAI says cached input-token reads get 90 percent discounts.

by read2 min views1 publishedSep 22, 2026

OpenAI today launched GPT–6 Sol and GPT–6 Luna, two new models that follow GPT–6 Astra. OpenAI says the new models were developed using methods similar to Astra's for improvements in professional work, factuality, coding, computer use, and alignment.

GPT–6 Sol is for everyday work, and it costs less than Astra. GPT–6 Luna is an even cheaper model. Luna is optimized for fast responses and higher volume, while Sol has more reasoning capability.

The new models outperform their predecessors on OpenAI's benchmark tests, and GPT–6 Sol makes "about half as many mistakes" as GPT–5.6 Sol. GPT–6 Sol shows improvements in coding and computer use compared to GPT–5.6 Sol, and it matches or beats Claude Fable 5.1 scores on some benchmarks.

OpenAI compared GPT–6 Sol to Opus 5 on several tests, but Anthropic released a new Opus 5.5 model today that outperforms Astra on some coding and knowledge work benchmarks.

GPT–6 Sol and Luna are adopting the improved communication style OpenAI introduced with Astra for simplified conversations with less jargon and fewer low-value details. Answers should be slightly shorter without losing substance, and the new models have lower rates of misleading claims about their coding work in OpenAI's alignment testing.

OpenAI is lowering token prices for Sol and Luna compared to their 5.6 counterparts, and prompt caching has been improved. OpenAI says agents will reuse more context, respond faster, and get 90 percent discounts on cached input-token reads.

GPT–6 Sol is priced at $2 per million input tokens and $10 per million output tokens, down by half. GPT–6 Luna costs $0.10 per million input tokens and $0.50 per million output tokens, down from $0.20 and $1.20, respectively. GPT–6 Sol costs half as much per token as Opus 5.5, but Anthropic says Opus 5.5 uses fewer tokens per task than Opus 5, so it's not a straight comparison.

GPT–6 Sol and GPT–6 Luna are rolling out in ChatGPT Work and Codex starting today for Plus, Pro, Business, Enterprise, and Edu users. GPT–6 Luna is available to Free and Go users in the desktop app. GPT–6 Sol and Luna are not yet available in Chat mode.

This article, "OpenAI's New GPT-6 Sol and Luna Models Bring Astra Improvements to Cheaper Tiers" first appeared on MacRumors.com

Discuss this article in our forums

── more in #large-language-models 4 stories · sorted by recency
── more on @openai 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/openai-s-new-gpt-6-s…] indexed:0 read:2min 2026-09-22 ·