Moondream 3 Preview vs NVIDIA Nemotron 3 Super
Side by side from live data: prices, limits, who serves each model and how reliably, and how they do on our chess arena.
In short
- Both cost about the same at a typical 3:1 input/output mix.
- Nemotron 3 Super accepts longer prompts: 262K vs 32K tokens.
- Nemotron 3 Super output costs $0.40 per 1M via DeepInfra, against the list price of $2.40.
- Moondream 3 Preview is the newer model (released 2026-10-15).
| Moondream 3 Preview | Nemotron 3 Super | |
|---|---|---|
| Vendor | Moondream | NVIDIA |
| Released | 2026-10-15 | 2026-03-11 |
| Input / 1M | $0.30 | $0.30 |
| Output / 1M | $2.50 | $2.40 |
| Cached input / 1M | — | $0.15 |
| Blended / 1M (3:1) | $0.85 | $0.825 |
| Context | 32K | 262K |
| Max output | 4K | 236K |
| Accepts | text, image | text |
| Reasoning | Yes | Yes |
| Providers on OpenRouter | 1 | 3 |
| Cheapest provider, output | $2.50 | $0.40 |
| Best provider uptime, 24 h | 100% | 100% |
| Chess arena rating | not playing | not playing |
| Vendor API status | — | — |
Cost calculator
Moondream 3 Preview
$55.50 per month$1.85 per day
Nemotron 3 Super
$54.00 per month$1.80 per day
Nemotron 3 Super saves $1.50 a month on this workload.
At list prices per 1M tokens; prompt caching and batch discounts are not included.
Compare Moondream 3 Preview with
Compare Nemotron 3 Super with