GPT-5.6 Luna (High) vs Hy3
Which is better for the money: GPT-5.6 Luna (High) by OpenAI or Hy3 by Tencent? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-25.
Clear winner
Verdict
Pick GPT-5.6 Luna (High)
Hy3 vs GPT-5.6 Luna (High): 21.2% less intelligence, 38.3% more cost to run the AA benchmark, and 174.1% slower end-to-end.
| Metric | GPT-5.6 Luna (High) | Hy3 | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 32.1 | 25.3 | −21.2% Hy3 is 21.2% less intelligent |
| Total AA cost (lower is better) | $108.15 | $149.61 | +38.3% Hy3 is 38.3% more expensive |
| End-to-end time (lower is better) | 12.29s | 33.69s | +174.1% Hy3 is 174.1% slower |
| Price per 1M tokens (in / out) | $0.20 / $1.20 | $0.14 / $0.56 | |
| Input modalities | Text, Image | Text |
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.
- Hy3 is off the frontier: at the same or lower cost, GPT-6 Luna (Max) scores 38.10 for $122.
GPT-5.6 Luna (High)Hy3Today'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/hy3 · Markdown: /api/v1/compare/gpt-5-6-luna-high/hy3?format=md · llms.txt · agent skill