Model comparison
DeepSeek-V2.5 (Sep 2024) vs Phi 3 Mini 4k Instruct June 2024
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 31.3 on the Noometry Index.
Last verified . 15 shared benchmarks.
Summary
- They share 15 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 7 categories and Phi 3 Mini 4k Instruct June 2024 in 1 category; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V2.5 (Sep 2024) leads 49.8 to 29.6.
Side by side
| DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 | |
|---|---|---|
| Provider | DeepSeek | Microsoft |
| Noometry Index | 37.6 | 31.3 |
| Released | 2024-09-06 | — |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 22 | 15 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), Phi 3 Mini 4k Instruct June 2024: 31.8 (#279)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 |
|---|---|---|
| LMArena Coding | 1309 | 1093 |
| Aider Polyglot | 17.8% | — |
| BigCodeBench Instruct | 48.6% | — |
| 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 Mini 4k Instruct June 2024: 20.8 (#231)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1087 |
Math DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), Phi 3 Mini 4k Instruct June 2024: 33.0 (#208)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 |
|---|---|---|
| LMArena Math | 1288 | 1152 |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), Phi 3 Mini 4k Instruct June 2024: 28.6 (#244)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 |
|---|---|---|
| LMArena Expert | 1266 | 1051 |
Multilingual DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), Phi 3 Mini 4k Instruct June 2024: 25.9 (#282)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 |
|---|---|---|
| LMArena Non-English | 1273 | 1013 |
| LMArena Chinese | 1318 | 1033 |
| LMArena German | 1258 | 1031 |
| LMArena Japanese | 1228 | 954 |
| LMArena Korean | 1209 | 880 |
| LMArena Russian | 1289 | 1019 |
| LMArena French | 1289 | — |
| LMArena Spanish | 1248 | — |
Instruction Following DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), Phi 3 Mini 4k Instruct June 2024: 54.1 (#282)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 |
|---|---|---|
| LMArena Instruction Following | 1280 | 1058 |
Long Context DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), Phi 3 Mini 4k Instruct June 2024: 31.7 (#277)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 |
|---|---|---|
| LMArena Longer Query | 1301 | 1042 |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), Phi 3 Mini 4k Instruct June 2024: 29.6 (#294)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Phi 3 Mini 4k Instruct June 2024 |
|---|---|---|
| LMArena Text | 1294 | 1080 |
| LMArena Creative Writing | 1285 | 1045 |
| LMArena Multi-Turn | 1297 | 1049 |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than Phi 3 Mini 4k Instruct June 2024?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 31.3 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or Phi 3 Mini 4k Instruct June 2024 better for coding?
They score almost the same on coding (31.7 vs 31.8); test both on your own repository before choosing.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Phi 3 Mini 4k Instruct June 2024 share?
15 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Phi 3 Mini 4k Instruct June 2024 has 15.