Model comparison
DeepSeek LLM 67B vs Phi-4
Phi-4 is the stronger model overall, scoring 31.2 to 24.9 on the Noometry Index.
Last verified . 15 shared benchmarks.
DeepSeek
MicrosoftSummary
- They share 15 benchmarks with published results for both. DeepSeek LLM 67B scores higher in 0 categories and Phi-4 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Phi-4 leads 32.6 to 7.0.
- The biggest single-benchmark swing is MATH Level 5: 6.4% for DeepSeek LLM 67B and 64.9% for Phi-4.
Side by side
| DeepSeek LLM 67B | Phi-4 | |
|---|---|---|
| Provider | DeepSeek | Microsoft |
| Noometry Index | 24.9 | 31.2 |
| Released | 2023-11-29 | 2024-12-11 |
| Weights | Open | Open |
| Context window | — | 128K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.07 |
| Output $ / M tokens | — | $0.14 |
| Results tracked | 15 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Phi-4 leads
DeepSeek LLM 67B: 31.9 (#278), Phi-4: 34.4 (#239)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| LMArena Coding | 1096 | 1231 |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
Agentic & Tool Use Not comparable
DeepSeek LLM 67B: —, Phi-4: 22.8 (#128)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
Reasoning Phi-4 leads
DeepSeek LLM 67B: 16.5 (#304), Phi-4: 17.7 (#291)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| Chess Puzzles | 0% | 1% |
| LMArena Hard Prompts | 1070 | 1220 |
| Epoch Capabilities Index | 110.5 | 130.42 |
| LiveBench Reasoning | — | 47.8% |
| LiveBench Data Analysis | — | 45.2% |
| LiveBench | — | 41.6% |
Math Phi-4 leads
DeepSeek LLM 67B: 8.7 (#324), Phi-4: 20.8 (#285)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 0.8% | 13.8% |
| LMArena Math | 1108 | 1246 |
| MATH Level 5 | 6.4% | 64.9% |
| LiveBench Math | — | 42% |
Knowledge Phi-4 leads
DeepSeek LLM 67B: 7.0 (#313), Phi-4: 32.6 (#209)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| GPQA Diamond | 24.6% | 56.1% |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| LMArena Expert | — | 1203 |
| MMLU | — | 84.8% |
Multilingual Phi-4 leads
DeepSeek LLM 67B: 29.4 (#267), Phi-4: 37.2 (#237)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| LMArena Non-English | 1073 | 1197 |
| LMArena Chinese | 1132 | 1212 |
| LMArena French | — | 1224 |
| LMArena German | — | 1222 |
| LMArena Japanese | — | 1158 |
| LMArena Korean | — | 1151 |
| LMArena Russian | — | 1209 |
| LMArena Spanish | — | 1234 |
Instruction Following Phi-4 leads
DeepSeek LLM 67B: 55.4 (#277), Phi-4: 60.4 (#251)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1079 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context Phi-4 leads
DeepSeek LLM 67B: 33.1 (#265), Phi-4: 36.9 (#226)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1092 | 1217 |
Writing & Preference Phi-4 leads
DeepSeek LLM 67B: 31.6 (#282), Phi-4: 40.5 (#244)
| Benchmark | DeepSeek LLM 67B | Phi-4 |
|---|---|---|
| LMArena Text | 1105 | 1217 |
| LMArena Creative Writing | 1067 | 1182 |
| LMArena Multi-Turn | 1082 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is DeepSeek LLM 67B better than Phi-4?
Phi-4 is the stronger model overall, scoring 31.2 to 24.9 on the Noometry Index.
Is DeepSeek LLM 67B or Phi-4 better for coding?
Phi-4 scores higher on coding benchmarks: 34.4 versus 31.9 in the Noometry coding category.
How many benchmarks do DeepSeek LLM 67B and Phi-4 share?
15 benchmarks have published results for both models. DeepSeek LLM 67B has 15 scored results on Noometry and Phi-4 has 37.