token.app › Compare › GPT-5.2 vs GPT-5.6 Luna
GPT-5.2 vs GPT-5.6 Luna
GPT-5.6 Luna leads on 5 of 5 shared benchmarks. GPT-5.6 Luna is 21× cheaper per token on a blended 3:1 basis. GPT-5.6 Luna accepts the larger context window (1.1M).
Head to head
Blended price uses a 3:1 input:output mix
| GPT-5.2 | GPT-5.6 Luna | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Input $/1M | $1.75 | $0.100 |
| Output $/1M | $14.00 | $0.600 |
| Blended $/1M | $4.81 | $0.225 |
| Context window | 400K | 1.1M |
| Max output | 128K | 128K |
| Released | Dec 2025 | Jul 2026 |
| Licence | Proprietary | Proprietary |
| GPQA Diamond | 91.4% | 91.6% |
| FrontierMath T1–3 | 67.4% | 82.1% |
| FrontierMath T4 | 31.7% | 61.0% |
| SimpleQA Verified | 38.9% | 41.7% |
| OTIS Mock AIME | 96.1% | 98.3% |
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose GPT-5.2 if…
- No clear advantage on the data we hold.
Choose GPT-5.6 Luna if…
- Costs less per token — $0.225 vs $4.81 blended
- Higher GPQA Diamond (91.6% vs 91.4%)
- Higher FrontierMath T1–3 (82.1% vs 67.4%)
- Higher FrontierMath T4 (61.0% vs 31.7%)
- Higher SimpleQA Verified (41.7% vs 38.9%)
- Larger context window (1.1M)
- Newer model (released Jul 2026)
FAQ
Which is better, GPT-5.2 or GPT-5.6 Luna?
Across the 5 benchmarks both models have independently-run scores for, GPT-5.2 leads on 0 and GPT-5.6 Luna leads on 5. Benchmarks are one input — price and context window are on this page too.
Is GPT-5.2 cheaper than GPT-5.6 Luna?
GPT-5.6 Luna is cheaper: $0.225 versus $4.81 per million tokens blended at 3:1 input:output. Input alone: $1.75 vs $0.100. Output alone: $14.00 vs $0.600.
What context windows do GPT-5.2 and GPT-5.6 Luna have?
GPT-5.2: 400,000 tokens. GPT-5.6 Luna: 1,050,000 tokens.
Who makes GPT-5.2 and GPT-5.6 Luna?
GPT-5.2 is made by OpenAI. GPT-5.6 Luna is made by OpenAI.