# GPT-5.6 Sol vs Phi-4

> GPT-5.6 Sol is the stronger model overall, scoring 65.0 to 31.2 on the Noometry Index. Phi-4 costs 91× less per token, which makes it the better buy when GPT-5.6 Sol's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gpt-5-6-sol-vs-phi-4
- Last updated: 2026-10-11
- Shared benchmarks: 23

## Summary

- They share 23 benchmarks with published results for both. GPT-5.6 Sol 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 GPT-5.6 Sol leads 85.6 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 100% for GPT-5.6 Sol and 13.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $4 / $20 for GPT-5.6 Sol.
- GPT-5.6 Sol accepts more context: 1.05M tokens versus 128K.
- Phi-4 has downloadable open weights; the other is API-only.

## Snapshot

| | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Noometry Index | 65.0 | 31.2 |
| Rank | 7 | 279 |
| Context | 1.05M | 128K |
| Input $/M | $4 | $0.07 |
| Output $/M | $20 | $0.14 |
| Weights | Proprietary | Open |

## Coding

- GPT-5.6 Sol: 65.1 (#7)
- Phi-4: 34.4 (#239)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| LMArena Coding | 1498 | 1231 |
| DeepSWE | 72.7% | — |
| FrontierCode | 47.5% | — |
| CursorBench | 41.7% | — |
| LMArena WebDev | 1618 | — |
| FrontierSWE | 32.2% | — |
| SciCode | 57.1% | — |
| GSO | 76.5% | — |
| WeirdML | 89.4% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| MirrorCode | 20% | — |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 2,177 | — |

## Agentic & Tool Use

- GPT-5.6 Sol: 50.3 (#7)
- Phi-4: 22.8 (#128)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| BALROG | 60% | 11.6% |
| APEX-Agents | 51.4% | — |
| Berkeley Function Calling Leaderboard | — | 28.8% |
| OSWorld 2.0 | 27.3% | — |
| τ²-bench Banking | 46.9% | — |
| PostTrainBench | 36.2% | — |
| GBAEval | 52.6% | — |
| GDP.pdf | 30.7% | — |
| LMArena Search | 1257 | — |
| Vending-Bench 2 | 9,619 | — |

## Reasoning

- GPT-5.6 Sol: 74.8 (#8)
- Phi-4: 17.7 (#291)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| Chess Puzzles | 64% | 1% |
| LMArena Hard Prompts | 1484 | 1220 |
| Epoch Capabilities Index | 161.66 | 130.42 |
| ARC-AGI-2 | 92.5% | — |
| SimpleBench | 71.7% | — |
| Kagi LLM Benchmark | 67% | — |
| NYT Connections (extended) | 93.8% | — |
| ARC-AGI-1 | 97.5% | — |
| CritPt | 32.3% | — |
| EnigmaEval | 37.1% | — |
| EBR-Bench | 44.8% | — |
| LiveBench Reasoning | — | 47.8% |
| Mystery Game Puzzles | 58% | — |
| DTBench | 96% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 59.2% | — |
| Surface Evolver Bench | 93.1% | — |
| Bench to the Future 3 | 0.14 | — |
| LiveBench | — | 41.6% |

## Math

- GPT-5.6 Sol: 85.6 (#9)
- Phi-4: 20.8 (#285)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 100% | 13.8% |
| LMArena Math | 1474 | 1246 |
| FrontierMath (Tiers 1-3) | 89.1% | — |
| FrontierMath Tier 4 | 82.9% | — |
| ProofBench | 83% | — |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
| FrontierMath Erdős | 0% | — |

## Knowledge

- GPT-5.6 Sol: 64.3 (#18)
- Phi-4: 32.6 (#209)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| GPQA Diamond | 93.5% | 56.1% |
| Vectara Hallucination Rate | 12.4% | 3.7% |
| LMArena Expert | 1516 | 1203 |
| SimpleQA Verified | 69.7% | — |
| Confabulations | — | 29.4% |
| MMLU | — | 84.8% |

## Multimodal

- GPT-5.6 Sol: 48.6 (#9)
- Phi-4: —

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| LMArena Vision | 1281 | — |
| Blueprint-Bench 2 | 33.6% | — |
| Furniture Assembly | 56.7% | — |
| LMArena Document | 1483 | — |

## Multilingual

- GPT-5.6 Sol: 55.3 (#32)
- Phi-4: 37.2 (#237)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| LMArena Non-English | 1452 | 1197 |
| LMArena Chinese | 1527 | 1212 |
| LMArena French | 1477 | 1224 |
| LMArena German | 1476 | 1222 |
| LMArena Japanese | 1471 | 1158 |
| LMArena Korean | 1442 | 1151 |
| LMArena Russian | 1468 | 1209 |
| LMArena Spanish | 1441 | 1234 |

## Instruction Following

- GPT-5.6 Sol: 77.7 (#16)
- Phi-4: 60.4 (#251)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1482 | 1201 |
| LiveBench Instruction Following | — | 58.4% |

## Long Context

- GPT-5.6 Sol: 45.4 (#42)
- Phi-4: 36.9 (#226)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1480 | 1217 |

## Writing & Preference

- GPT-5.6 Sol: 73.3 (#12)
- Phi-4: 40.5 (#244)

| Benchmark | GPT-5.6 Sol | Phi-4 |
|---|---|---|
| LMArena Text | 1457 | 1217 |
| LMArena Creative Writing | 1448 | 1182 |
| LMArena Multi-Turn | 1460 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| EQ-Bench Creative Writing | 1972 | — |
| EQ-Bench 4 | 1250 | — |
| LiveBench Language | — | 25.6% |

## FAQ

### Is GPT-5.6 Sol better than Phi-4?

GPT-5.6 Sol is the stronger model overall, scoring 65.0 to 31.2 on the Noometry Index. Phi-4 costs 91× less per token, which makes it the better buy when GPT-5.6 Sol's lead doesn't matter for your workload.

### Which is cheaper, GPT-5.6 Sol or Phi-4?

Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; GPT-5.6 Sol lists at $4 and $20.

### Is GPT-5.6 Sol or Phi-4 better for coding?

GPT-5.6 Sol scores higher on coding benchmarks: 65.1 versus 34.4 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.6 Sol does, with 1.05M tokens against 128K.

### How many benchmarks do GPT-5.6 Sol and Phi-4 share?

23 benchmarks have published results for both models. GPT-5.6 Sol has 65 scored results on Noometry and Phi-4 has 37.