Model comparison
Mercury 2 vs Phi-4
Mercury 2 is the stronger model overall, scoring 39.1 to 31.2 on the Noometry Index. Phi-4 costs 4.3× less per token, which makes it the better buy when Mercury 2's lead doesn't matter for your workload.
Last verified . 12 shared benchmarks.
MicrosoftSummary
- They share 12 benchmarks with published results for both. Mercury 2 scores higher in 6 categories and Phi-4 in 1 category; 6 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Mercury 2 leads 53.8 to 40.5.
- The biggest single-benchmark swing is Vectara Hallucination Rate: 12.3% for Mercury 2 and 3.7% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.25 / $0.75 for Mercury 2.
- Phi-4 has downloadable open weights; the other is API-only.
Side by side
| Mercury 2 | Phi-4 | |
|---|---|---|
| Provider | Inception | Microsoft |
| Noometry Index | 39.1 | 31.2 |
| Released | 2026-02-20 | 2024-12-11 |
| Weights | Proprietary | Open |
| Context window | 128K | 128K |
| Max output | 50K | 4K |
| Input $ / M tokens | $0.25 | $0.07 |
| Output $ / M tokens | $0.75 | $0.14 |
| Results tracked | 17 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Mercury 2: 33.5 (#255), Phi-4: 34.4 (#239)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| LMArena Coding | 1391 | 1231 |
| LMArena WebDev | 1171 | — |
| SciCode | 38.7% | — |
| WeirdML | 43.2% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 785.58 | — |
Agentic & Tool Use Not comparable
Mercury 2: —, Phi-4: 22.8 (#128)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
Reasoning Mercury 2 leads
Mercury 2: 23.8 (#170), Phi-4: 17.7 (#291)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1362 | 1220 |
| CritPt | 0.8% | — |
| Chess Puzzles | — | 1% |
| LiveBench Reasoning | — | 47.8% |
| LiveBench Data Analysis | — | 45.2% |
| Epoch Capabilities Index | — | 130.42 |
| LiveBench | — | 41.6% |
Math Not comparable
Mercury 2: —, Phi-4: 20.8 (#285)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 13.8% |
| LiveBench Math | — | 42% |
| LMArena Math | — | 1246 |
| MATH Level 5 | — | 64.9% |
Knowledge Mercury 2 leads
Mercury 2: 36.2 (#172), Phi-4: 32.6 (#209)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| Vectara Hallucination Rate | 12.3% | 3.7% |
| LMArena Expert | 1358 | 1203 |
| GPQA Diamond | — | 56.1% |
| Confabulations | — | 29.4% |
| MMLU | — | 84.8% |
Multilingual Mercury 2 leads
Mercury 2: 46.6 (#157), Phi-4: 37.2 (#237)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| LMArena Non-English | 1331 | 1197 |
| LMArena Chinese | 1417 | 1212 |
| LMArena Russian | 1304 | 1209 |
| LMArena French | — | 1224 |
| LMArena German | — | 1222 |
| LMArena Japanese | — | 1158 |
| LMArena Korean | — | 1151 |
| LMArena Spanish | — | 1234 |
Instruction Following Mercury 2 leads
Mercury 2: 70.2 (#165), Phi-4: 60.4 (#251)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1329 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context Mercury 2 leads
Mercury 2: 40.5 (#154), Phi-4: 36.9 (#226)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1330 | 1217 |
Writing & Preference Mercury 2 leads
Mercury 2: 53.8 (#155), Phi-4: 40.5 (#244)
| Benchmark | Mercury 2 | Phi-4 |
|---|---|---|
| LMArena Text | 1355 | 1217 |
| LMArena Creative Writing | 1289 | 1182 |
| LMArena Multi-Turn | 1358 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is Mercury 2 better than Phi-4?
Mercury 2 is the stronger model overall, scoring 39.1 to 31.2 on the Noometry Index. Phi-4 costs 4.3× less per token, which makes it the better buy when Mercury 2's lead doesn't matter for your workload.
Which is cheaper, Mercury 2 or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Mercury 2 lists at $0.25 and $0.75.
Is Mercury 2 or Phi-4 better for coding?
They score almost the same on coding (33.5 vs 34.4); test both on your own repository before choosing.
Which has the bigger context window?
Both accept 128K tokens.
How many benchmarks do Mercury 2 and Phi-4 share?
12 benchmarks have published results for both models. Mercury 2 has 17 scored results on Noometry and Phi-4 has 37.