Updated

OpenAI
GPT 5.6 Luna
OpenAI's fastest, most affordable GPT-5.6 model for quick chat and high-volume work.
What is GPT 5.6 Luna?
GPT 5.6 Luna is the fastest and most affordable model in OpenAI's GPT-5.6 series, generally available since July 9, 2026.
OpenAI built it for cost-sensitive, high-volume work and says it roughly fills the nano tier of earlier GPT-5 families. OpenAI is rolling it out as the default ChatGPT model for Free and Go users. Like the rest of the series, it takes text and images in a 1.05M-token context window. GPT-6 Luna, Super Agent's default model, is its newer counterpart.
- Provider
- OpenAI
- Context window
- 1.05M tokens
- Input
- Text, Image
- Reasoning
- Off to max
- Max output
- 128K tokens
- Knowledge cutoff
- Feb 16, 2026
- Released
- Jul 9, 2026
How GPT 5.6 Luna performs
Published by OpenAI
Results OpenAI reported for GPT 5.6 Luna. Each lab runs its own setup, so compare these with other models only inside the same table.
| Benchmark | Score | Setting |
|---|---|---|
| Agents' Last Exam | 50.3% | |
| SWE-Bench Pro | 62.7% | |
| DeepSWE v1.1 | 67.2% | |
| Terminal-Bench 2.1 | 84.7% | |
| OSWorld 2.0 | 45.6% | |
| BrowseComp | 83.3% | |
| GPQA Diamond | 92.3% | |
| MMMU Pro | 78.4% | no tools |
| Toolathlon | 53.4% | |
| OpenAI MRCR v2, 8 needles | 41.3% | 512K to 1M tokens |
Source: openai.com
On LMArena
People compare two anonymous answers and vote for the better one. Ranked here against the other chat models you can pick in Super Agent.
Text
#24 of 27 in Super Agent
All text prompts, with style control on. #75 of 418 on the full leaderboard, from 41,426 votes.
- 1Claude Opus 5.51509
- 2Claude Fable 51503
- 3Claude Fable 5.11503
- 4Claude Opus 51491
- 5Gemini 3.7 Flash1490
- 24GPT 5.6 Luna1453
Coding
#23 of 27 in Super Agent
Text prompts that ask for code. #76 of 413 on the full leaderboard, from 11,215 votes.
- 1Claude Fable 51551
- 2Claude Opus 5.51548
- 3GPT-6 Astra1543
- 4Kimi K31539
- 5Claude Fable 5.11538
- 23GPT 5.6 Luna1502
WebDev
#22 of 27 in Super Agent
Building working web apps in Code Arena. #56 of 149 on the full leaderboard, from 12,570 votes.
- 1Claude Opus 5.51807
- 2GPT-6 Astra1775
- 3Claude Sonnet 5.51768
- 4GPT-6.1 Sol1749
- 5Claude Fable 5.11743
- 22GPT 5.6 Luna1519
Math
#18 of 27 in Super Agent
Text prompts with math problems. #42 of 404 on the full leaderboard, from 1,946 votes.
- 1Claude Fable 5.11529
- 2Claude Fable 51521
- 3Claude Opus 51517
- 4GPT-6.1 Sol1513
- 5Claude Opus 5.51512
- 18GPT 5.6 Luna1478
Hard prompts
#23 of 27 in Super Agent
Complex, multi-part and demanding text prompts. #76 of 418 on the full leaderboard, from 27,208 votes.
- 1Claude Opus 5.51535
- 2Claude Fable 51529
- 3Claude Fable 5.11523
- 4Kimi K31516
- 5Claude Opus 51515
- 23GPT 5.6 Luna1475
Creative writing
#26 of 27 in Super Agent
Stories, poems and other original writing. #92 of 416 on the full leaderboard, from 8,692 votes.
- 1Claude Opus 5.51520
- 2Claude Fable 51501
- 3Gemini 3.7 Flash1492
- 4Claude Fable 5.11487
- 5Gemini 3.1 Pro Preview1481
- 26GPT 5.6 Luna1410
Vision
#21 of 23 in Super Agent
Prompts that include images, with style control on. #48 of 167 on the full leaderboard, from 11,055 votes.
- 1Claude Fable 51309
- 2Gemini 3.7 Flash1295
- 3GPT-6.1 Sol1294
- 4Claude Opus 5.51293
- 5Claude Opus 51288
- 21GPT 5.6 Luna1257
Source: LMArena, leaderboard published . Scores move as new votes come in.
What you can do with GPT 5.6 Luna
Fastest, lowest-cost tier of the GPT-5.6 series
Becoming the default ChatGPT model for Free and Go users
Reasoning off for instant answers, up to max when needed
Same 1.05M-token window and image input as the rest of GPT-5.6
How to prompt GPT 5.6 Luna
- Keep requests short and specific for the fastest answers.
- Turn reasoning off when you just need variations or rewrites.
- For harder tasks, try GPT 5.6 Terra or GPT 5.6 Sol.
Where GPT 5.6 Luna fits best
Quick chat
Sorting and labeling lists
Short prompt variations
Simple agent steps
How GPT 5.6 Luna compares
| Model | LMArena text | LMArena coding | Context window | Reasoning |
|---|---|---|---|---|
| GPT 5.6 Luna | 1453#75 | 1502#76 | 1.05M tokens | Off to max |
| Gemini 3.5 Flash Lite | 1455#71 | 1502#77 | 1M tokens | Low, medium, high |
| Grok 4.6 | 1456#66 | 1510#60 | 500K tokens | Low to extra high |
| DeepSeek V4 Pro | 1458#64 | 1501#79 | 1M tokens | Off, medium, high, extra high |
| GPT-6 Sol | 1459#62 | 1519#38 | 1.05M tokens | Off to max |
Good to know before you start
- GPT-6 Luna is the newer light model from OpenAI
- Not built for the hardest reasoning work
- Text replies only; ask Super Agent for an image or clip and a Flashloop generation model renders it
Start chatting with GPT 5.6 Luna
Chat with GPT 5.6 Luna in Super AgentSelect GPT 5.6 Luna
Open Super Agent on Flashloop and choose GPT 5.6 Luna in the model picker.
Enter your prompt
Ask for what you need: an answer, a plan, or the images and videos you want the agent to make.
Send and keep going
Send the message. GPT 5.6 Luna answers in Super Agent and runs any generations you asked for.
Questions about GPT 5.6 Luna
What is GPT 5.6 Luna used for?
OpenAI built it for cost-sensitive, high-volume work and is rolling it out as ChatGPT's default model for Free and Go users.
How does GPT 5.6 Luna compare with GPT-6 Luna?
GPT-6 Luna is OpenAI's newer light model and the default in Super Agent. GPT 5.6 Luna is the previous generation's equivalent.
Which reasoning levels does GPT 5.6 Luna support?
Off, low, medium, high, extra high and max. Super Agent starts it at low.
Where is GPT 5.6 Luna in the model picker?
Under All models.
More chat models on Flashloop
Gemini 3.5 Flash LiteGoogle's fastest, lowest-cost Gemini 3.5 model for extraction, routing and subagent tasks.
Grok 4.6xAI's Grok 4.6 for turning a broad product idea into a working first version, with 500K context.
DeepSeek V4 ProDeepSeek's largest V4 model for coding, knowledge and STEM; Super Agent runs the April preview.
GPT-6 SolOpenAI's GPT-6 model between Luna and Astra, trained like Astra and able to reply with reasoning off.