Model comparison
Llama 4 Maverick vs Phi-4
Llama 4 Maverick and Phi-4 score almost the same on the Noometry Index (30.9 vs 31.2), so choose on price, context window or the category you care about most.
Last verified . 27 shared benchmarks.
Meta
MicrosoftSummary
- They share 27 benchmarks with published results for both. Llama 4 Maverick scores higher in 5 categories and Phi-4 in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in instruction following, where Llama 4 Maverick leads 71.7 to 60.4.
- The biggest single-benchmark swing is GPQA Diamond: 67% for Llama 4 Maverick and 56.1% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.19 / $0.65 for Llama 4 Maverick.
Side by side
| Llama 4 Maverick | Phi-4 | |
|---|---|---|
| Provider | Meta | Microsoft |
| Noometry Index | 30.9 | 31.2 |
| Released | 2025-04-05 | 2024-12-11 |
| Weights | Open | Open |
| Context window | 128K | 128K |
| Max output | 4K | 4K |
| Input $ / M tokens | $0.19 | $0.07 |
| Output $ / M tokens | $0.65 | $0.14 |
| Results tracked | 54 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Phi-4 leads
Llama 4 Maverick: 26.6 (#324), Phi-4: 34.4 (#239)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| BigCodeBench Instruct | 49.7% | 45.5% |
| LMArena Coding | 1302 | 1231 |
| BigCodeBench Complete | 61.4% | 55.4% |
| SWE-bench Verified (bash only) | 21% | — |
| Aider Polyglot | 15.6% | — |
| SciCode | 33.1% | — |
| WeirdML | 24.5% | — |
| LiveBench Coding | — | 30.7% |
| ALE-Bench | 172.97 | — |
Agentic & Tool Use Llama 4 Maverick leads
Llama 4 Maverick: 28.2 (#91), Phi-4: 22.8 (#128)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 37.3% | 28.8% |
| BALROG | — | 11.6% |
Reasoning Phi-4 leads
Llama 4 Maverick: 10.1 (#342), Phi-4: 17.7 (#291)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1281 | 1220 |
| Epoch Capabilities Index | 132.2 | 130.42 |
| ARC-AGI-2 | 0% | — |
| SimpleBench | 27.7% | — |
| Kagi LLM Benchmark | 55.9% | — |
| NYT Connections (extended) | 8% | — |
| ARC-AGI-1 | 4.4% | — |
| CritPt | 0% | — |
| Chess Puzzles | — | 1% |
| EnigmaEval | 0.6% | — |
| LiveBench Reasoning | — | 47.8% |
| DTBench | 61.9% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 15.9% | — |
| ForecastBench | 57.5 | — |
| LiveBench | — | 41.6% |
Math Llama 4 Maverick leads
Llama 4 Maverick: 26.0 (#262), Phi-4: 20.8 (#285)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 20.6% | 13.8% |
| LMArena Math | 1299 | 1246 |
| MATH Level 5 | 73% | 64.9% |
| Omni-MATH | 42.2% | — |
| LiveBench Math | — | 42% |
| FrontierMath (Feb 2025 set) | 0.7% | — |
Knowledge Too close to call
Llama 4 Maverick: 33.4 (#204), Phi-4: 32.6 (#209)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| GPQA Diamond | 67% | 56.1% |
| Confabulations | 22.6% | 29.4% |
| Vectara Hallucination Rate | 8.2% | 3.7% |
| LMArena Expert | 1259 | 1203 |
| Humanity's Last Exam | 5.7% | — |
| MMLU-Pro | 81% | — |
| GPQA (HELM) | 65% | — |
| MMLU | — | 84.8% |
Multimodal Not comparable
Llama 4 Maverick: 31.6 (#105), Phi-4: —
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| LMArena Vision | 1142 | — |
| GeoBench | 52% | — |
| SpatialViz-Bench | 31.8% | — |
Multilingual Llama 4 Maverick leads
Llama 4 Maverick: 42.2 (#195), Phi-4: 37.2 (#237)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| LMArena Non-English | 1269 | 1197 |
| LMArena Chinese | 1277 | 1212 |
| LMArena French | 1259 | 1224 |
| LMArena German | 1291 | 1222 |
| LMArena Japanese | 1207 | 1158 |
| LMArena Korean | 1203 | 1151 |
| LMArena Russian | 1286 | 1209 |
| LMArena Spanish | 1293 | 1234 |
Instruction Following Llama 4 Maverick leads
Llama 4 Maverick: 71.7 (#146), Phi-4: 60.4 (#251)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1267 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
| IFEval | 90.8% | — |
Long Context Phi-4 leads
Llama 4 Maverick: 31.4 (#279), Phi-4: 36.9 (#226)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1280 | 1217 |
| Fiction.LiveBench | 46.2% | — |
Writing & Preference Phi-4 leads
Llama 4 Maverick: 38.8 (#252), Phi-4: 40.5 (#244)
| Benchmark | Llama 4 Maverick | Phi-4 |
|---|---|---|
| LMArena Text | 1287 | 1217 |
| LMArena Creative Writing | 1267 | 1182 |
| Short-Story Creative Writing | 62% | 62.6% |
| LMArena Multi-Turn | 1289 | 1206 |
| EQ-Bench Creative Writing | 860 | — |
| WildBench | 80% | — |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is Llama 4 Maverick better than Phi-4?
Llama 4 Maverick and Phi-4 score almost the same on the Noometry Index (30.9 vs 31.2), so choose on price, context window or the category you care about most.
Which is cheaper, Llama 4 Maverick or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Llama 4 Maverick lists at $0.19 and $0.65.
Is Llama 4 Maverick or Phi-4 better for coding?
Phi-4 scores higher on coding benchmarks: 34.4 versus 26.6 in the Noometry coding category.
Which has the bigger context window?
Both accept 128K tokens.
How many benchmarks do Llama 4 Maverick and Phi-4 share?
27 benchmarks have published results for both models. Llama 4 Maverick has 54 scored results on Noometry and Phi-4 has 37.