GPT-5.6 Luna (High) vs GPT-5.6 Sol (Low)
Which is better for the money: GPT-5.6 Luna (High) by OpenAI or GPT-5.6 Sol (Low) by OpenAI? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-27.
Trade-off
Verdict
Depends on your budget
GPT-5.6 Sol (Low) scores 1.4 points higher (33.5 vs 32.1) but costs 5.9× as much to run the benchmark ($637 vs $108; premium vs mid-range cost classes). They are not alternatives at one budget: pick GPT-5.6 Sol (Low) when you need intelligence above 32.1, and GPT-5.6 Luna (High) when 32.1 is enough.
| Metric | GPT-5.6 Luna (High) | GPT-5.6 Sol (Low) | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 32.1 | 33.5 | +4.4% GPT-5.6 Sol (Low) is 4.4% more intelligent |
| Total AA cost (lower is better) | $108.15 | $637.28 | +489.3% GPT-5.6 Sol (Low) is 5.9× more expensive |
| End-to-end time (lower is better) | 16.30s | 8.38s | −48.6% GPT-5.6 Sol (Low) is 48.6% faster |
| Price per 1M tokens (in / out) | $0.20 / $1.20 | $4.00 / $20.0 | |
| Input modalities | Text, Image | Text, Image |
GPT-5.6 Sol (Low) vs GPT-5.6 Luna (High): 4.4% more intelligence, 489.3% more cost to run the AA benchmark, and 48.6% faster end-to-end.
Today's Pareto frontier
Against the frontier
- GPT-5.6 Luna (High) is off the frontier: at the same or lower cost, Claude Haiku 5.5 (High) scores 37.80 for $94.1.
- GPT-5.6 Sol (Low) is off the frontier: at the same or lower cost, GPT-6.1 Sol (High) scores 50.20 for $521.
GPT-5.6 Luna (High)GPT-5.6 Sol (Low)Today's Pareto frontierEvery other model
Matchup
Compare something else
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/gpt-5-6-luna-high/gpt-5-6-sol-low · Markdown: /api/v1/compare/gpt-5-6-luna-high/gpt-5-6-sol-low?format=md · llms.txt · agent skill