Model comparison
Gemma 4 26B A4B IT vs Phi-4
Gemma 4 26B A4B IT is the stronger model overall, scoring 43.5 to 31.2 on the Noometry Index.
Last verified . 19 shared benchmarks.
Google
MicrosoftSummary
- They share 19 benchmarks with published results for both. Gemma 4 26B A4B IT scores higher in 8 categories and Phi-4 in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemma 4 26B A4B IT leads 47.6 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 82.2% for Gemma 4 26B A4B IT and 13.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.0675 / $0.23 for Gemma 4 26B A4B IT.
- Gemma 4 26B A4B IT accepts more context: 262K tokens versus 128K.
Side by side
| Gemma 4 26B A4B IT | Phi-4 | |
|---|---|---|
| Provider | Microsoft | |
| Noometry Index | 43.5 | 31.2 |
| Released | 2026-04-02 | 2024-12-11 |
| Weights | Open | Open |
| Context window | 262K | 128K |
| Max output | 33K | 4K |
| Input $ / M tokens | $0.0675 | $0.07 |
| Output $ / M tokens | $0.23 | $0.14 |
| Results tracked | 28 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 39.0 (#164), Phi-4: 34.4 (#239)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| LMArena Coding | 1447 | 1231 |
| LMArena WebDev | 1359 | — |
| SciCode | 40% | — |
| WeirdML | 35.2% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 927.17 | — |
Agentic & Tool Use Not comparable
Gemma 4 26B A4B IT: —, Phi-4: 22.8 (#128)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
Reasoning Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 21.8 (#213), Phi-4: 17.7 (#291)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| Chess Puzzles | 6% | 1% |
| LMArena Hard Prompts | 1439 | 1220 |
| Epoch Capabilities Index | 141.85 | 130.42 |
| CritPt | 0% | — |
| LiveBench Reasoning | — | 47.8% |
| DTBench | 74.9% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 29.7% | — |
| LiveBench | — | 41.6% |
Math Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 47.6 (#67), Phi-4: 20.8 (#285)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 82.2% | 13.8% |
| LMArena Math | 1470 | 1246 |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
Knowledge Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 45.7 (#85), Phi-4: 32.6 (#209)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| GPQA Diamond | 73.2% | 56.1% |
| Vectara Hallucination Rate | 5.2% | 3.7% |
| LMArena Expert | 1447 | 1203 |
| Confabulations | — | 29.4% |
| MMLU | — | 84.8% |
Multimodal Not comparable
Gemma 4 26B A4B IT: 40.6 (#46), Phi-4: —
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| LMArena Vision | 1260 | — |
Multilingual Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 53.1 (#71), Phi-4: 37.2 (#237)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| LMArena Non-English | 1421 | 1197 |
| LMArena Chinese | 1495 | 1212 |
| LMArena French | 1460 | 1224 |
| LMArena Russian | 1434 | 1209 |
| LMArena Spanish | 1417 | 1234 |
| LMArena German | — | 1222 |
| LMArena Japanese | — | 1158 |
| LMArena Korean | — | 1151 |
Instruction Following Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 74.8 (#82), Phi-4: 60.4 (#251)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1420 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 43.6 (#91), Phi-4: 36.9 (#226)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1428 | 1217 |
Writing & Preference Gemma 4 26B A4B IT leads
Gemma 4 26B A4B IT: 58.6 (#115), Phi-4: 40.5 (#244)
| Benchmark | Gemma 4 26B A4B IT | Phi-4 |
|---|---|---|
| LMArena Text | 1434 | 1217 |
| LMArena Creative Writing | 1402 | 1182 |
| LMArena Multi-Turn | 1441 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| EQ-Bench Creative Writing | 1305 | — |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is Gemma 4 26B A4B IT better than Phi-4?
Gemma 4 26B A4B IT is the stronger model overall, scoring 43.5 to 31.2 on the Noometry Index.
Which is cheaper, Gemma 4 26B A4B IT or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Gemma 4 26B A4B IT lists at $0.0675 and $0.23.
Is Gemma 4 26B A4B IT or Phi-4 better for coding?
Gemma 4 26B A4B IT scores higher on coding benchmarks: 39.0 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
Gemma 4 26B A4B IT does, with 262K tokens against 128K.
How many benchmarks do Gemma 4 26B A4B IT and Phi-4 share?
19 benchmarks have published results for both models. Gemma 4 26B A4B IT has 28 scored results on Noometry and Phi-4 has 37.