Opens in a new tab

News

OfficialSource: OpenAI announcements for GPT-6 Sol and Luna (Sep 22, 2026) and GPT-6.1 Sol (Sep 29, 2026), plus OpenAI's API pricing page.

GPT-6.1 Sol and GPT-6 Luna: What OpenAI’s Cheaper Models Mean for You

OpenAI shipped GPT-6 Sol and Luna in September, then upgraded Sol to GPT-6.1 a week later. Here's what you actually get in ChatGPT, Codex and the API.

A phone showing the OpenAI logo on top of a laptop
Photo: Levart_Photographer / Unsplash
On this page
  1. API prices
  2. What changed in GPT-6.1 Sol
  3. Related

DGTULS may earn a commission when you purchase through links on our site. This never affects our ratings or which products we recommend.

OpenAI’s GPT-6 family now has three tiers below the price of its top model: GPT-6 Astra (most capable), GPT-6.1 Sol (the middle tier) and GPT-6 Luna (the cheapest). Sol and Luna arrived on September 22, 2026. A week later, at DevDay on September 29, OpenAI replaced Sol with an upgraded GPT-6.1 Sol at the same price (OpenAI).

API prices

Prices per 1 million tokens, checked on October 23, 2026 (OpenAI pricing).

Model API ID Input Cached input Output
GPT-6 Astra gpt-6-astra $10.00 $1.00 $50.00
GPT-6.1 Sol gpt-6.1-sol $2.00 $0.10 $10.00
GPT-6 Luna gpt-6-luna $0.10 $0.01 $0.50

When Sol and Luna launched, OpenAI said they cost half of the GPT-5.6 promotional prices (Sol was $4/$20 and Luna $0.20/$1.20 per million tokens on GPT-5.6).

What changed in GPT-6.1 Sol

According to OpenAI’s announcement (these are OpenAI’s own tests; we haven’t verified them):

  • Fewer factual errors: the error rate on OpenAI’s factuality test fell from 11.4% to 7.7% at low reasoning effort.
  • Coding: on DeepSWE v1.1, OpenAI says 6.1 Sol matches GPT-6 Astra at about one-fifth of the cost.
  • Agent tasks: OpenAI reports fewer cases of agents ignoring explicit restrictions or hiding broken tools.

Independent tests point the same way, with one trade-off. DeepLearning.AI reports that GPT-6.1 Sol scored 52 on the Artificial Analysis Intelligence Index, against 48 for GPT-6 Sol, but generates tokens more slowly (about 58 vs 88 tokens per second). It also drops GPT-6 Sol’s option to turn reasoning off (The Batch).

Related

Prices checked on October 23, 2026.

Sources

Essential cookies keep DGTULS working and are always on. Optional cookies help us measure traffic and show ads. Accept, reject or choose which ones to allow. Learn more in our Cookie Policy, Privacy Policy and Terms of Use.