Model comparison
GPT-5.4 nano vs Phi-4
GPT-5.4 nano is the stronger model overall, scoring 41.9 to 31.2 on the Noometry Index. Phi-4 costs 5.3× less per token, which makes it the better buy when GPT-5.4 nano's lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
MicrosoftSummary
- They share 22 benchmarks with published results for both. GPT-5.4 nano 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 GPT-5.4 nano leads 40.9 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 87.8% for GPT-5.4 nano and 13.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.20 / $1.25 for GPT-5.4 nano.
- GPT-5.4 nano accepts more context: 400K tokens versus 128K.
- Phi-4 has downloadable open weights; the other is API-only.
Side by side
| GPT-5.4 nano | Phi-4 | |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Noometry Index | 41.9 | 31.2 |
| Released | 2026-03-17 | 2024-12-11 |
| Weights | Proprietary | Open |
| Context window | 400K | 128K |
| Max output | 128K | 4K |
| Input $ / M tokens | $0.20 | $0.07 |
| Output $ / M tokens | $1.25 | $0.14 |
| Results tracked | 40 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 nano leads
GPT-5.4 nano: 43.6 (#84), Phi-4: 34.4 (#239)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| LMArena Coding | 1405 | 1231 |
| SciCode | 46.9% | — |
| WeirdML | 49.2% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 1,005 | — |
Agentic & Tool Use Not comparable
GPT-5.4 nano: —, Phi-4: 22.8 (#128)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
Reasoning GPT-5.4 nano leads
GPT-5.4 nano: 23.7 (#173), Phi-4: 17.7 (#291)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| Chess Puzzles | 30% | 1% |
| LMArena Hard Prompts | 1381 | 1220 |
| Epoch Capabilities Index | 145.81 | 130.42 |
| ARC-AGI-2 | 5.7% | — |
| Kagi LLM Benchmark | 39.7% | — |
| ARC-AGI-1 | 51.5% | — |
| CritPt | 9.3% | — |
| LiveBench Reasoning | — | 47.8% |
| Mystery Game Puzzles | 9% | — |
| DTBench | 80.3% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 36.9% | — |
| ForecastBench | 57.3 | — |
| LiveBench | — | 41.6% |
Math GPT-5.4 nano leads
GPT-5.4 nano: 40.9 (#88), Phi-4: 20.8 (#285)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 87.8% | 13.8% |
| LMArena Math | 1406 | 1246 |
| FrontierMath (Tiers 1-3) | 44.9% | — |
| FrontierMath Tier 4 | 12.2% | — |
| ProofBench | 5% | — |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
| FrontierMath (Feb 2025 set) | 25.9% | — |
| FrontierMath Tier 4 (v1) | 6.3% | — |
Knowledge GPT-5.4 nano leads
GPT-5.4 nano: 41.9 (#103), Phi-4: 32.6 (#209)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| GPQA Diamond | 78.5% | 56.1% |
| Vectara Hallucination Rate | 3.1% | 3.7% |
| LMArena Expert | 1396 | 1203 |
| SimpleQA Verified | 11.7% | — |
| Confabulations | — | 29.4% |
| MMLU | — | 84.8% |
Multimodal Not comparable
GPT-5.4 nano: 36.7 (#78), Phi-4: —
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| LMArena Vision | 1196 | — |
Multilingual GPT-5.4 nano leads
GPT-5.4 nano: 48.6 (#140), Phi-4: 37.2 (#237)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| LMArena Non-English | 1359 | 1197 |
| LMArena Chinese | 1392 | 1212 |
| LMArena French | 1396 | 1224 |
| LMArena German | 1367 | 1222 |
| LMArena Japanese | 1343 | 1158 |
| LMArena Korean | 1320 | 1151 |
| LMArena Russian | 1363 | 1209 |
| LMArena Spanish | 1371 | 1234 |
Instruction Following GPT-5.4 nano leads
GPT-5.4 nano: 71.9 (#144), Phi-4: 60.4 (#251)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1362 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context GPT-5.4 nano leads
GPT-5.4 nano: 41.6 (#137), Phi-4: 36.9 (#226)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1366 | 1217 |
Writing & Preference GPT-5.4 nano leads
GPT-5.4 nano: 55.7 (#142), Phi-4: 40.5 (#244)
| Benchmark | GPT-5.4 nano | Phi-4 |
|---|---|---|
| LMArena Text | 1372 | 1217 |
| LMArena Creative Writing | 1314 | 1182 |
| LMArena Multi-Turn | 1382 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is GPT-5.4 nano better than Phi-4?
GPT-5.4 nano is the stronger model overall, scoring 41.9 to 31.2 on the Noometry Index. Phi-4 costs 5.3× less per token, which makes it the better buy when GPT-5.4 nano's lead doesn't matter for your workload.
Which is cheaper, GPT-5.4 nano or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; GPT-5.4 nano lists at $0.20 and $1.25.
Is GPT-5.4 nano or Phi-4 better for coding?
GPT-5.4 nano scores higher on coding benchmarks: 43.6 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 nano does, with 400K tokens against 128K.
How many benchmarks do GPT-5.4 nano and Phi-4 share?
22 benchmarks have published results for both models. GPT-5.4 nano has 40 scored results on Noometry and Phi-4 has 37.