Model comparison
DeepSeek-V2.5 (Sep 2024) vs phi-3-medium 14B
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 29.7 on the Noometry Index.
Last verified . 2 shared benchmarks.
DeepSeek
MicrosoftSummary
- They share 2 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 2 categories and phi-3-medium 14B in 1 category; 3 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V2.5 (Sep 2024) leads 34.8 to 9.1.
- The biggest single-benchmark swing is BigCodeBench Instruct: 48.6% for DeepSeek-V2.5 (Sep 2024) and 37.6% for phi-3-medium 14B.
Side by side
| DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B | |
|---|---|---|
| Provider | DeepSeek | Microsoft |
| Noometry Index | 37.6 | 29.7 |
| Released | 2024-09-06 | 2024-04-23 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 22 | 13 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding phi-3-medium 14B leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), phi-3-medium 14B: 36.8 (#201)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B |
|---|---|---|
| BigCodeBench Instruct | 48.6% | 37.6% |
| BigCodeBench Complete | 53.2% | 48.7% |
| Aider Polyglot | 17.8% | — |
| LMArena Coding | 1309 | — |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Reasoning Not comparable
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), phi-3-medium 14B: —
| Benchmark | DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B |
|---|---|---|
| LMArena Hard Prompts | 1289 | — |
| Adversarial NLI | — | 55.8% |
| BIG-Bench Hard | — | 81.4% |
| Epoch Capabilities Index | — | 121.23 |
| HellaSwag | — | 82.4% |
| WinoGrande | — | 81.5% |
Math DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), phi-3-medium 14B: 27.3 (#250)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B |
|---|---|---|
| LMArena Math | 1288 | — |
| MATH Level 5 | — | 17.6% |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), phi-3-medium 14B: 9.1 (#306)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B |
|---|---|---|
| GPQA Diamond | — | 27.6% |
| LMArena Expert | 1266 | — |
| ARC (AI2) Challenge | — | 91.6% |
| MMLU | — | 78% |
| OpenBookQA | — | 87.4% |
| TriviaQA | — | 73.9% |
Multilingual Not comparable
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), phi-3-medium 14B: —
| Benchmark | DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B |
|---|---|---|
| LMArena Non-English | 1273 | — |
| LMArena Chinese | 1318 | — |
| LMArena French | 1289 | — |
| LMArena German | 1258 | — |
| LMArena Japanese | 1228 | — |
| LMArena Korean | 1209 | — |
| LMArena Russian | 1289 | — |
| LMArena Spanish | 1248 | — |
Instruction Following Not comparable
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), phi-3-medium 14B: —
| Benchmark | DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B |
|---|---|---|
| LMArena Instruction Following | 1280 | — |
Long Context Not comparable
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), phi-3-medium 14B: —
| Benchmark | DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B |
|---|---|---|
| LMArena Longer Query | 1301 | — |
Writing & Preference Not comparable
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), phi-3-medium 14B: —
| Benchmark | DeepSeek-V2.5 (Sep 2024) | phi-3-medium 14B |
|---|---|---|
| LMArena Text | 1294 | — |
| LMArena Creative Writing | 1285 | — |
| LMArena Multi-Turn | 1297 | — |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than phi-3-medium 14B?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 29.7 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or phi-3-medium 14B better for coding?
phi-3-medium 14B scores higher on coding benchmarks: 36.8 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and phi-3-medium 14B share?
2 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and phi-3-medium 14B has 13.