Model comparison
Mistral Medium 3.5 vs Phi-4
Mistral Medium 3.5 is the stronger model overall, scoring 40.2 to 31.2 on the Noometry Index. Phi-4 costs 34× less per token, which makes it the better buy when Mistral Medium 3.5's lead doesn't matter for your workload.
Last verified . 17 shared benchmarks.
Mistral AI
MicrosoftSummary
- They share 17 benchmarks with published results for both. Mistral Medium 3.5 scores higher in 7 categories and Phi-4 in 1 category; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Mistral Medium 3.5 leads 39.1 to 20.8.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $1.50 / $7.50 for Mistral Medium 3.5.
- Mistral Medium 3.5 accepts more context: 262K tokens versus 128K.
Side by side
| Mistral Medium 3.5 | Phi-4 | |
|---|---|---|
| Provider | Mistral AI | Microsoft |
| Noometry Index | 40.2 | 31.2 |
| Released | — | 2024-12-11 |
| Weights | Open | Open |
| Context window | 262K | 128K |
| Max output | 210K | 4K |
| Input $ / M tokens | $1.50 | $0.07 |
| Output $ / M tokens | $7.50 | $0.14 |
| Results tracked | 22 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mistral Medium 3.5 leads
Mistral Medium 3.5: 36.0 (#213), Phi-4: 34.4 (#239)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Coding | 1461 | 1231 |
| LMArena WebDev | 1264 | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
Agentic & Tool Use Not comparable
Mistral Medium 3.5: —, Phi-4: 22.8 (#128)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
Reasoning Too close to call
Mistral Medium 3.5: 17.3 (#295), Phi-4: 17.7 (#291)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1436 | 1220 |
| Epoch Capabilities Index | 141.35 | 130.42 |
| Kagi LLM Benchmark | 41.4% | — |
| NYT Connections (extended) | 12.9% | — |
| Chess Puzzles | — | 1% |
| LiveBench Reasoning | — | 47.8% |
| LiveBench Data Analysis | — | 45.2% |
| LiveBench | — | 41.6% |
Math Mistral Medium 3.5 leads
Mistral Medium 3.5: 39.1 (#113), Phi-4: 20.8 (#285)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Math | 1431 | 1246 |
| OTIS Mock AIME 2024-2025 | — | 13.8% |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
Knowledge Mistral Medium 3.5 leads
Mistral Medium 3.5: 40.0 (#126), Phi-4: 32.6 (#209)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Expert | 1432 | 1203 |
| GPQA Diamond | — | 56.1% |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| MMLU | — | 84.8% |
Multimodal Not comparable
Mistral Medium 3.5: 38.3 (#65), Phi-4: —
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Vision | 1223 | — |
Multilingual Mistral Medium 3.5 leads
Mistral Medium 3.5: 51.9 (#100), Phi-4: 37.2 (#237)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Non-English | 1404 | 1197 |
| LMArena Chinese | 1442 | 1212 |
| LMArena French | 1448 | 1224 |
| LMArena German | 1451 | 1222 |
| LMArena Korean | 1385 | 1151 |
| LMArena Russian | 1395 | 1209 |
| LMArena Spanish | 1409 | 1234 |
| LMArena Japanese | — | 1158 |
Instruction Following Mistral Medium 3.5 leads
Mistral Medium 3.5: 74.6 (#90), Phi-4: 60.4 (#251)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1415 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context Mistral Medium 3.5 leads
Mistral Medium 3.5: 43.2 (#103), Phi-4: 36.9 (#226)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1415 | 1217 |
Writing & Preference Mistral Medium 3.5 leads
Mistral Medium 3.5: 58.5 (#117), Phi-4: 40.5 (#244)
| Benchmark | Mistral Medium 3.5 | Phi-4 |
|---|---|---|
| LMArena Text | 1421 | 1217 |
| LMArena Creative Writing | 1374 | 1182 |
| LMArena Multi-Turn | 1423 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| EQ-Bench 4 | 993 | — |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is Mistral Medium 3.5 better than Phi-4?
Mistral Medium 3.5 is the stronger model overall, scoring 40.2 to 31.2 on the Noometry Index. Phi-4 costs 34× less per token, which makes it the better buy when Mistral Medium 3.5's lead doesn't matter for your workload.
Which is cheaper, Mistral Medium 3.5 or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Mistral Medium 3.5 lists at $1.50 and $7.50.
Is Mistral Medium 3.5 or Phi-4 better for coding?
Mistral Medium 3.5 scores higher on coding benchmarks: 36.0 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
Mistral Medium 3.5 does, with 262K tokens against 128K.
How many benchmarks do Mistral Medium 3.5 and Phi-4 share?
17 benchmarks have published results for both models. Mistral Medium 3.5 has 22 scored results on Noometry and Phi-4 has 37.