Model comparison
o4-mini vs Phi-4
o4-mini is the stronger model overall, scoring 41.6 to 31.2 on the Noometry Index. Phi-4 costs 22× less per token, which makes it the better buy when o4-mini's lead doesn't matter for your workload.
Last verified . 26 shared benchmarks.
MicrosoftSummary
- They share 26 benchmarks with published results for both. o4-mini scores higher in 9 categories and Phi-4 in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where o4-mini leads 40.8 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 81.7% for o4-mini and 13.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $1.10 / $4.40 for o4-mini.
- o4-mini accepts more context: 200K tokens versus 128K.
- Phi-4 has downloadable open weights; the other is API-only.
Side by side
| o4-mini | Phi-4 | |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Noometry Index | 41.6 | 31.2 |
| Released | 2025-04-16 | 2024-12-11 |
| Weights | Proprietary | Open |
| Context window | 200K | 128K |
| Max output | 100K | 4K |
| Input $ / M tokens | $1.10 | $0.07 |
| Output $ / M tokens | $4.40 | $0.14 |
| Results tracked | 60 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o4-mini leads
o4-mini: 40.9 (#127), Phi-4: 34.4 (#239)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| LMArena Coding | 1368 | 1231 |
| SWE-bench Verified (bash only) | 45% | — |
| Aider Polyglot | 72% | — |
| GSO | 3.6% | — |
| WeirdML | 52.6% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| CadEval | 62% | — |
| ALE-Bench | 826.17 | — |
| AlgoTune | 1.72 | — |
Agentic & Tool Use o4-mini leads
o4-mini: 32.6 (#61), Phi-4: 22.8 (#128)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 53.2% | 28.8% |
| GDPval | 25.3% | — |
| BALROG | — | 11.6% |
| METR Time Horizons | 63.9% | — |
Reasoning o4-mini leads
o4-mini: 24.6 (#162), Phi-4: 17.7 (#291)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| Chess Puzzles | 26% | 1% |
| LMArena Hard Prompts | 1351 | 1220 |
| Epoch Capabilities Index | 145.64 | 130.42 |
| ARC-AGI-2 | 6.1% | — |
| SimpleBench | 38.7% | — |
| Kagi LLM Benchmark | 67.6% | — |
| ARC-AGI-1 | 58.7% | — |
| CritPt | 0.6% | — |
| EnigmaEval | 9.2% | — |
| LiveBench Reasoning | — | 47.8% |
| Mystery Game Puzzles | 5% | — |
| DTBench | 77.6% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 26.5% | — |
| ForecastBench | 61.8 | — |
| LiveBench | — | 41.6% |
Math o4-mini leads
o4-mini: 40.8 (#89), Phi-4: 20.8 (#285)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 81.7% | 13.8% |
| LMArena Math | 1389 | 1246 |
| MATH Level 5 | 97.8% | 64.9% |
| FrontierMath (Tiers 1-3) | 36.1% | — |
| FrontierMath Tier 4 | 4.9% | — |
| Omni-MATH | 72% | — |
| LiveBench Math | — | 42% |
| FrontierMath (Feb 2025 set) | 24.8% | — |
| FrontierMath Tier 4 (v1) | 6.3% | — |
Knowledge o4-mini leads
o4-mini: 43.6 (#91), Phi-4: 32.6 (#209)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| GPQA Diamond | 79.6% | 56.1% |
| Confabulations | 15.8% | 29.4% |
| Vectara Hallucination Rate | 18.6% | 3.7% |
| LMArena Expert | 1343 | 1203 |
| Humanity's Last Exam | 18.1% | — |
| SimpleQA Verified | 19.6% | — |
| MMLU-Pro | 82% | — |
| GPQA (HELM) | 73.5% | — |
| MMLU | — | 84.8% |
Multimodal Not comparable
o4-mini: 40.2 (#49), Phi-4: —
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| LMArena Vision | 1194 | — |
| GeoBench | 64% | — |
| VPCT | 57.5% | — |
Multilingual o4-mini leads
o4-mini: 47.0 (#154), Phi-4: 37.2 (#237)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| LMArena Non-English | 1337 | 1197 |
| LMArena Chinese | 1354 | 1212 |
| LMArena French | 1364 | 1224 |
| LMArena German | 1336 | 1222 |
| LMArena Japanese | 1308 | 1158 |
| LMArena Korean | 1312 | 1151 |
| LMArena Russian | 1334 | 1209 |
| LMArena Spanish | 1347 | 1234 |
Instruction Following o4-mini leads
o4-mini: 75.2 (#68), Phi-4: 60.4 (#251)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1321 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
| IFEval | 92.8% | — |
Long Context o4-mini leads
o4-mini: 45.5 (#33), Phi-4: 36.9 (#226)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1315 | 1217 |
| Fiction.LiveBench | 77.8% | — |
Writing & Preference o4-mini leads
o4-mini: 54.0 (#152), Phi-4: 40.5 (#244)
| Benchmark | o4-mini | Phi-4 |
|---|---|---|
| LMArena Text | 1353 | 1217 |
| LMArena Creative Writing | 1294 | 1182 |
| Short-Story Creative Writing | 75% | 62.6% |
| LMArena Multi-Turn | 1350 | 1206 |
| WildBench | 85.4% | — |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is o4-mini better than Phi-4?
o4-mini is the stronger model overall, scoring 41.6 to 31.2 on the Noometry Index. Phi-4 costs 22× less per token, which makes it the better buy when o4-mini's lead doesn't matter for your workload.
Which is cheaper, o4-mini or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; o4-mini lists at $1.10 and $4.40.
Is o4-mini or Phi-4 better for coding?
o4-mini scores higher on coding benchmarks: 40.9 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
o4-mini does, with 200K tokens against 128K.
How many benchmarks do o4-mini and Phi-4 share?
26 benchmarks have published results for both models. o4-mini has 60 scored results on Noometry and Phi-4 has 37.