DeepSeek-V4-Flash vs Llama-3.3 Nemotron Super 49B v1
DeepSeek vs NVIDIA — benchmarks, pricing, and capabilities side by side.
- •DeepSeek-V4-Flash has the higher intelligence index (89.4 vs 63.9)
- •Llama-3.3 Nemotron Super 49B v1 is cheaper ($0.00 vs $0.10 per 1M input)
| DeepSeek-V4-Flash | Llama-3.3 Nemotron Super 49B v1 | |
|---|---|---|
| Intelligence index | 89.4 | 63.9 |
| Developer | DeepSeek | NVIDIA |
| Type | LLM | LLM |
| Access | Open weights | Open weights |
| Context window | 1,048,576 tokens | — |
| Input price | $0.10 / 1M | $0.00 / 1M |
| Output price | $0.20 / 1M | $0.00 / 1M |
| Speed | 109 tok/s | — |
| Released | April 24, 2026 | March 18, 2025 |
| Parameters | 284B (13B active) | 49900000000 |
| Input modalities | Text | — |
| Output modalities | Text | — |
Shared benchmarks
DeepSeek-V4-Flash
Llama-3.3 Nemotron Super 49B v1
GPQA Diamond
89.4
66.7
Humanity’s Last Exam
32.1
6.5