Model comparison
Phi-4 vs Qwen2.5 7B Instruct
Phi-4 is the stronger model overall, scoring 31.2 to 29.0 on the Noometry Index.
Last verified . 8 shared benchmarks.
Microsoft
Alibaba (Qwen)Summary
- They share 8 benchmarks with published results for both. Phi-4 scores higher in 3 categories and Qwen2.5 7B Instruct in 4 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Phi-4 leads 32.6 to 17.0.
- The biggest single-benchmark swing is GPQA Diamond: 56.1% for Phi-4 and 35.5% for Qwen2.5 7B Instruct.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.17 / $0.70 for Qwen2.5 7B Instruct.
- Qwen2.5 7B Instruct accepts more context: 131K tokens versus 128K.
Side by side
| Phi-4 | Qwen2.5 7B Instruct | |
|---|---|---|
| Provider | Microsoft | Alibaba (Qwen) |
| Noometry Index | 31.2 | 29.0 |
| Released | 2024-12-11 | 2024-09 |
| Weights | Open | Open |
| Context window | 128K | 131K |
| Max output | 4K | 8K |
| Input $ / M tokens | $0.07 | $0.17 |
| Output $ / M tokens | $0.14 | $0.70 |
| Results tracked | 37 | 15 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen2.5 7B Instruct leads
Phi-4: 34.4 (#239), Qwen2.5 7B Instruct: 36.5 (#208)
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| BigCodeBench Instruct | 45.5% | 37.6% |
| BigCodeBench Complete | 55.4% | 46.1% |
| LiveBench Coding | 30.7% | — |
| LMArena Coding | 1231 | — |
Agentic & Tool Use Qwen2.5 7B Instruct leads
Phi-4: 22.8 (#128), Qwen2.5 7B Instruct: 23.8 (#124)
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| BALROG | 11.6% | 7.8% |
| Berkeley Function Calling Leaderboard | 28.8% | — |
Reasoning Phi-4 leads
Phi-4: 17.7 (#291), Qwen2.5 7B Instruct: 14.8 (#322)
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| Chess Puzzles | 1% | 0% |
| Epoch Capabilities Index | 130.42 | 118.51 |
| LiveBench Reasoning | 47.8% | — |
| LMArena Hard Prompts | 1220 | — |
| DTBench | — | 47.7% |
| LiveBench Data Analysis | 45.2% | — |
| LMCA | — | 6.4% |
| LiveBench | 41.6% | — |
Math Phi-4 leads
Phi-4: 20.8 (#285), Qwen2.5 7B Instruct: 12.6 (#306)
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 13.8% | 2.5% |
| Omni-MATH | — | 29.4% |
| LiveBench Math | 42% | — |
| LMArena Math | 1246 | — |
| MATH Level 5 | 64.9% | — |
Knowledge Phi-4 leads
Phi-4: 32.6 (#209), Qwen2.5 7B Instruct: 17.0 (#286)
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| GPQA Diamond | 56.1% | 35.5% |
| MMLU | 84.8% | 72.9% |
| MMLU-Pro | — | 53.9% |
| Confabulations | 29.4% | — |
| Vectara Hallucination Rate | 3.7% | — |
| GPQA (HELM) | — | 34.1% |
| LMArena Expert | 1203 | — |
Multilingual Not comparable
Phi-4: 37.2 (#237), Qwen2.5 7B Instruct: —
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Non-English | 1197 | — |
| LMArena Chinese | 1212 | — |
| LMArena French | 1224 | — |
| LMArena German | 1222 | — |
| LMArena Japanese | 1158 | — |
| LMArena Korean | 1151 | — |
| LMArena Russian | 1209 | — |
| LMArena Spanish | 1234 | — |
Instruction Following Qwen2.5 7B Instruct leads
Phi-4: 60.4 (#251), Qwen2.5 7B Instruct: 63.2 (#231)
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| LiveBench Instruction Following | 58.4% | — |
| IFEval | — | 74.1% |
| LMArena Instruction Following | 1201 | — |
Long Context Not comparable
Phi-4: 36.9 (#226), Qwen2.5 7B Instruct: —
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Longer Query | 1217 | — |
Writing & Preference Qwen2.5 7B Instruct leads
Phi-4: 40.5 (#244), Qwen2.5 7B Instruct: 48.8 (#195)
| Benchmark | Phi-4 | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Text | 1217 | — |
| LMArena Creative Writing | 1182 | — |
| Short-Story Creative Writing | 62.6% | — |
| WildBench | — | 73.1% |
| LMArena Multi-Turn | 1206 | — |
| LiveBench Language | 25.6% | — |
Frequently asked questions
Is Phi-4 better than Qwen2.5 7B Instruct?
Phi-4 is the stronger model overall, scoring 31.2 to 29.0 on the Noometry Index.
Which is cheaper, Phi-4 or Qwen2.5 7B Instruct?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Qwen2.5 7B Instruct lists at $0.17 and $0.70.
Is Phi-4 or Qwen2.5 7B Instruct better for coding?
Qwen2.5 7B Instruct scores higher on coding benchmarks: 36.5 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
Qwen2.5 7B Instruct does, with 131K tokens against 128K.
How many benchmarks do Phi-4 and Qwen2.5 7B Instruct share?
8 benchmarks have published results for both models. Phi-4 has 37 scored results on Noometry and Qwen2.5 7B Instruct has 15.