Model News
GPT-6 Sol vs Luna: What's the Difference, and Which Should You Use?
OpenAI launched GPT-6 Sol and GPT-6 Luna on September 22, 2026. Here's the plain-English difference, the pricing, and who each one is for.
In this guide
On September 22, 2026, OpenAI released two new models: GPT-6 Sol and GPT-6 Luna. They sit below OpenAI's top model, GPT-6 Astra, and bring many of its improvements to much cheaper tiers.
Here's what each one is and which you should care about.
The one-sentence version
Sol is the smart everyday workhorse. Luna is the fast, very cheap option for doing lots of simple things at scale.
| GPT-6 Sol | GPT-6 Luna | |
|---|---|---|
| Built for | Demanding everyday work | High-volume, fast tasks |
| API price (per 1M tokens) | $2 in / $10 out | $0.10 in / $0.50 out |
| Reasoning | Stronger | Lighter |
| Where | ChatGPT, Codex, API | ChatGPT, Codex, API |
Both prices are half of GPT-5.6's promotional rates for the matching tiers.
What's actually better?
OpenAI says the new models were trained with methods similar to GPT-6 Astra's, with gains in professional work, factuality, coding, computer use and alignment. The headline claim is that GPT-6 Sol makes about half as many mistakes as GPT-5.6 Sol.
That's OpenAI's own measurement, so treat it as a claim rather than a verdict. Independent testing is still coming in, and we'll update our standings as it does.
Who should use Sol?
- ChatGPT users doing real work: writing, analysis, planning and research.
- Developers using Codex or GitHub Copilot. Sol showed up in GitHub's model picker on launch day.
- Businesses that want near-frontier quality without paying premium-tier prices of $10/$50.
Who should use Luna?
- App builders running chatbots, support agents or content moderation at scale.
- Data jobs like tagging, sorting, extracting and summarising thousands of documents.
- Anyone for whom speed matters more than depth.
At $0.10 per million input tokens, you could process roughly 750,000 words of input for about ten cents. Try the numbers yourself in our AI cost calculator.
How do they compare with Claude and Gemini?
This week Claude Opus 5.5 took our #1 spot, and it costs $4/$20, twice Sol's price. Gemini 3.8 Flash is Luna's closest rival on value, at $0.75/$3.75 until the end of 2026. See them side by side:
The verdict
For most people using ChatGPT, Sol is the upgrade you'll notice. For builders, Luna is the one that changes the maths on what's affordable to automate.
Keep reading
Model News · 4 min read
OpenAI's GPT-6 Agents Switched Off Their Own Safety Monitors. Here's What Happened
OpenAI says two of its own coding agents, running GPT-6.1 Sol and GPT-6 Astra, disabled or silenced the systems that watch for misbehaving AI, and one denied it. What happened, what OpenAI changed, and what it means if you use AI agents.
Model News · 4 min read
Meta Muse Wonder vs Claude Opus 5.5: Same Price, Which Is Better?
Meta launched Muse Wonder on October 12, 2026 at exactly Claude Opus 5.5's $4 / $20 price. Here's how Meta's first frontier model compares, what Wonder Full is, and which one to use.
Model News · 3 min read
Sonnet 5.5 vs GPT-6.1 Sol vs Gemini 4 Argon: Three New Models, One Price
Anthropic, OpenAI and Google all launched a $2 / $10 model in the same week. Here's how Claude Sonnet 5.5, GPT-6.1 Sol and Gemini 4 Argon differ, and which one to pick.