DeepSeekvsPhi
Comparing DeepSeek and Microsoft. Average Podium Score across top models: DeepSeek (71.6) vs Phi (58.4). Flagship showdown: Deepseek V4 Pro High Preview (73.0) vs Phi 3 Medium 4k Instruct (60.0).
DeepSeek (DeepSeek)
High-efficiency, cost-effective reasoning and coding open-weight models.
Phi (Microsoft)
Highly capable Small Language Models (SLMs) from Microsoft Research.
Top Models Showdown
| Rank | Model | Lab | Score | Coding | Reasoning | Speed | Price / 1M |
|---|---|---|---|---|---|---|---|
| #1 | Deepseek V4 Pro High Preview | DeepSeek | 73.0 | — | — | 50 t/s | $1.1/M |
| #2 | Deepseek V4 Flash High Preview | DeepSeek | 72.0 | — | — | 50 t/s | $1.1/M |
| #3 | Deepseek V3.2 Exp Thinking | DeepSeek | 71.0 | — | — | 50 t/s | $1.1/M |
| #4 | Deepseek V3.2 Thinking | DeepSeek | 71.0 | — | — | 50 t/s | $1.1/M |
| #5 | Deepseek V3.2 Exp | DeepSeek | 71.0 | — | — | 50 t/s | $1.1/M |
| #6 | Phi 3 Medium 4k Instruct | Microsoft | 60.0 | — | — | 50 t/s | $3/M |
| #7 | Wizardlm 70b | Microsoft | 59.0 | — | — | 50 t/s | $3/M |
| #8 | Phi 3 Small 8k Instruct | Microsoft | 59.0 | — | — | 50 t/s | $3/M |
| #9 | Wizardlm 13b | Microsoft | 57.0 | — | — | 50 t/s | $3/M |
| #10 | Phi 3 Mini 4k Instruct June 2024 | Microsoft | 57.0 | — | — | 50 t/s | $3/M |
Looking for more ecosystem comparisons? DeepSeek vs Claude · DeepSeek vs ChatGPT / GPT · DeepSeek vs Kimi · DeepSeek vs Gemini · DeepSeek vs Qwen · DeepSeek vs Grok