Phi 4 vs DeepSeek V4 Flash 0731 (Thinking)
DeepSeek V4 Flash 0731 (Thinking) is cheaper on output tokens, while DeepSeek V4 Flash 0731 (Thinking) offers a larger context window. Choose Phi 4 or DeepSeek V4 Flash 0731 (Thinking) based on the trade-off between cost, context, and the benchmarks that matter for your use case.
| Spec | Phi 4 | DeepSeek V4 Flash 0731 (Thinking) |
|---|---|---|
| Provider | Microsoft | NanoGPT |
| Input / 1M tokens | $0.07 | $0.04 |
| Output / 1M tokens | $0.14 | $0.13 |
| Context window | 16K | 1M |
| Parameters | — | — |
| Open weights | — | Yes |
| Released | — | Jul 2026 |
| GPQA Diamond score | 56.1 | — |
| Epoch Capabilities Index score | 131 | — |
| LMArena (Chatbot Arena) Elo score | 1217 | — |
FAQ
Is Phi 4 or DeepSeek V4 Flash 0731 (Thinking) cheaper?
DeepSeek V4 Flash 0731 (Thinking) is cheaper on output tokens ($0.13 vs $0.14 per 1M).
Which has the larger context window, Phi 4 or DeepSeek V4 Flash 0731 (Thinking)?
DeepSeek V4 Flash 0731 (Thinking) has the larger context window (1M tokens).
Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.