GPT-6 Luna vs Nemotron 3 Super
GPT-6 Luna is ahead on some work, and they tie on the rest. GPT-6 Luna is the cheaper one, by 2.8×.
GrindBench-PubSec
CGPT-6 LunaOpenAI54% 49–60%0.11¢ a task · 300 tasksDNemotron 3 SuperNVIDIA40% 34–45%0.15¢ a task · 300 tasks
GPT-6 Luna is ahead here by 15 points, more than chance explains.
| Kind of work | GPT-6 Luna | Nemotron 3 Super |
|---|---|---|
| ATIP redaction | 2% | 2% |
| Procurement evaluation | 69% | 81% |
| Performance analysis | 54% | 20% |
| Bilingual equivalence | 90% | 75% |
GrindBench-Banking
FGPT-6 LunaOpenAI18% 10–28%0.38¢ a task · 60 tasksFNemotron 3 SuperNVIDIA5% 0–12%1.2¢ a task · 60 tasks
Too close to call here: their ranges overlap, so on another set of tasks the order could flip.
| Kind of work | GPT-6 Luna | Nemotron 3 Super |
|---|---|---|
| Lending & mortgages | 20% | 10% |
| AML & KYC | 20% | 10% |
| Deposits & payments | 20% | 0% |
| Complaints & conduct | 30% | 10% |
| Credit risk | 10% | 0% |
| Capital & liquidity | 10% | 0% |
Each kind of work here rests on 10 tasks per model, so a single row can easily be 20 points off. A lead of 10 points or more is in bold; read it as a hint, not a result.