DeepSeek-V4-Flash vs K2-V2
DeepSeek vs MBZUAI Institute of Foundation Models — benchmarks, pricing, and capabilities side by side.
- •DeepSeek-V4-Flash has the higher intelligence index (89.4 vs 73.6)
- •K2-V2 is cheaper ($0.00 vs $0.10 per 1M input)
| DeepSeek-V4-Flash | K2-V2 | |
|---|---|---|
| Intelligence index | 89.4 | 73.6 |
| Developer | DeepSeek | MBZUAI Institute of Foundation Models |
| Type | LLM | LLM |
| Access | 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 | December 5, 2025 |
| Parameters | 284B (13B active) | — |
| Input modalities | Text | — |
| Output modalities | Text | — |
Shared benchmarks
DeepSeek-V4-Flash
K2-V2
GPQA Diamond
89.4
68.1
Humanity’s Last Exam
32.1
9.8