GPT-5.6 Luna (High) vs gpt-oss-20b (High)
Which is better for the money: GPT-5.6 Luna (High) by OpenAI or gpt-oss-20b (High) by OpenAI? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-26.
Trade-off
Verdict
Depends on your budget
GPT-5.6 Luna (High) scores 23.1 points higher (32.1 vs 9.0) but costs 3.8× as much to run the benchmark ($108 vs $28.8; mid-range vs budget cost classes). They are not alternatives at one budget: pick GPT-5.6 Luna (High) when you need intelligence above 9.0, and gpt-oss-20b (High) when 9.0 is enough.
| Metric | GPT-5.6 Luna (High) | gpt-oss-20b (High) | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 32.1 | 9.0 | −72.0% gpt-oss-20b (High) is 72.0% less intelligent |
| Total AA cost (lower is better) | $108.15 | $28.81 | −73.4% gpt-oss-20b (High) is 73.4% cheaper |
| End-to-end time (lower is better) | 14.62s | 14.42s | −1.4% gpt-oss-20b (High) is 1.4% faster |
| Price per 1M tokens (in / out) | $0.20 / $1.20 | $0.05 / $0.22 | |
| Input modalities | Text, Image | Text |
gpt-oss-20b (High) vs GPT-5.6 Luna (High): 72.0% less intelligence, 73.4% less cost to run the AA benchmark, and 1.4% 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-oss-20b (High) is off the frontier: at the same or lower cost, GPT-6 Luna (Low) scores 21.50 for $10.6.
GPT-5.6 Luna (High)gpt-oss-20b (High)Today's Pareto frontierOpen-weights 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-oss-20b · Markdown: /api/v1/compare/gpt-5-6-luna-high/gpt-oss-20b?format=md · llms.txt · agent skill