Model comparison
DeepSeek-V2.5 (Sep 2024) vs Phi 3 Small 8k Instruct
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 29.3 on the Noometry Index.
Last verified . 17 shared benchmarks.
DeepSeek
MicrosoftSummary
- They share 17 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 8 categories and Phi 3 Small 8k Instruct in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V2.5 (Sep 2024) leads 49.8 to 31.1.
Side by side
| DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct | |
|---|---|---|
| Provider | DeepSeek | Microsoft |
| Noometry Index | 37.6 | 29.3 |
| Released | 2024-09-06 | 2024-04-23 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 22 | 32 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), Phi 3 Small 8k Instruct: 27.9 (#318)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Coding | 1309 | 1101 |
| Aider Polyglot | 17.8% | — |
| BigCodeBench Instruct | 48.6% | — |
| LiveBench Coding | — | 20.3% |
| BigCodeBench Complete | 53.2% | — |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), Phi 3 Small 8k Instruct: 14.8 (#323)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1100 |
| LiveBench Reasoning | — | 15.9% |
| LiveBench Data Analysis | — | 30.3% |
| Adversarial NLI | — | 58.1% |
| BIG-Bench Hard | — | 79.1% |
| HellaSwag | — | 77% |
| LiveBench | — | 24% |
| WinoGrande | — | 81.5% |
Math DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), Phi 3 Small 8k Instruct: 27.6 (#248)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Math | 1288 | 1151 |
| LiveBench Math | — | 17.6% |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), Phi 3 Small 8k Instruct: 29.1 (#240)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Expert | 1266 | 1067 |
| ARC (AI2) Challenge | — | 90.7% |
| MMLU | — | 75.7% |
| OpenBookQA | — | 88% |
| TriviaQA | — | 58.1% |
Multilingual DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), Phi 3 Small 8k Instruct: 28.5 (#272)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Non-English | 1273 | 1058 |
| LMArena Chinese | 1318 | 1061 |
| LMArena French | 1289 | 1135 |
| LMArena German | 1258 | 1080 |
| LMArena Japanese | 1228 | 966 |
| LMArena Korean | 1209 | 894 |
| LMArena Russian | 1289 | 1111 |
| LMArena Spanish | 1248 | 1111 |
Instruction Following DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), Phi 3 Small 8k Instruct: 51.9 (#292)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Instruction Following | 1280 | 1087 |
| LiveBench Instruction Following | — | 47.2% |
Long Context DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), Phi 3 Small 8k Instruct: 33.0 (#267)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Longer Query | 1301 | 1088 |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), Phi 3 Small 8k Instruct: 31.1 (#284)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Text | 1294 | 1110 |
| LMArena Creative Writing | 1285 | 1083 |
| LMArena Multi-Turn | 1297 | 1068 |
| LiveBench Language | — | 12.9% |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than Phi 3 Small 8k Instruct?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 29.3 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or Phi 3 Small 8k Instruct better for coding?
DeepSeek-V2.5 (Sep 2024) scores higher on coding benchmarks: 31.7 versus 27.9 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Phi 3 Small 8k Instruct share?
17 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Phi 3 Small 8k Instruct has 32.