# Phi-4 vs Qwen3.6 35B-A3B

> Qwen3.6 35B-A3B is the stronger model overall, scoring 37.2 to 30.9 on the Noometry Index. Phi-4 costs 6.4× less per token, which makes it the better buy when Qwen3.6 35B-A3B's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/phi-4-vs-qwen3-6-35b-a3b
- Last updated: 2026-10-10
- Shared benchmarks: 4

## Summary

- They share 4 benchmarks with published results for both. Phi-4 scores higher in 1 category and Qwen3.6 35B-A3B in 4 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Qwen3.6 35B-A3B leads 51.4 to 32.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 13.8% for Phi-4 and 86.7% for Qwen3.6 35B-A3B.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.25 / $1.49 for Qwen3.6 35B-A3B.
- Qwen3.6 35B-A3B accepts more context: 262K tokens versus 128K.

## Snapshot

| | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| Provider | Microsoft | Alibaba (Qwen) |
| Noometry Index | 30.9 | 37.2 |
| Rank | 269 | 194 |
| Context | 128K | 262K |
| Input $/M | $0.07 | $0.25 |
| Output $/M | $0.14 | $1.49 |
| Weights | Open | Open |

## Coding

- Phi-4: 33.8 (#237)
- Qwen3.6 35B-A3B: 36.7 (#190)

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| SciCode | — | 35.8% |
| WeirdML | — | 34.5% |
| BigCodeBench Instruct | 45.5% | — |
| LiveBench Coding | 30.7% | — |
| LMArena Coding | 1231 | — |
| BigCodeBench Complete | 55.4% | — |

## Agentic & Tool Use

- Phi-4: 22.6 (#120)
- Qwen3.6 35B-A3B: 21.9 (#126)

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| Terminal-Bench | — | 23% |
| Berkeley Function Calling Leaderboard | 28.8% | — |
| BALROG | 11.6% | — |

## Reasoning

- Phi-4: 17.0 (#287)
- Qwen3.6 35B-A3B: 27.2 (#112)

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| Chess Puzzles | 1% | 26% |
| Epoch Capabilities Index | 130.43 | 143.92 |
| NYT Connections (extended) | — | 41.6% |
| CritPt | — | 0.3% |
| LiveBench Reasoning | 47.8% | — |
| LMArena Hard Prompts | 1220 | — |
| Mystery Game Puzzles | — | 22% |
| DTBench | — | 73.9% |
| LiveBench Data Analysis | 45.2% | — |
| LMCA | — | 29.7% |
| Surface Evolver Bench | — | 44.4% |
| LiveBench | 41.6% | — |

## Math

- Phi-4: 19.6 (#279)
- Qwen3.6 35B-A3B: 37.9 (#90)

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 13.8% | 86.7% |
| FrontierMath (Tiers 1-3) | — | 20.4% |
| LiveBench Math | 42% | — |
| LMArena Math | 1246 | — |
| MATH Level 5 | 64.9% | — |

## Knowledge

- Phi-4: 32.5 (#204)
- Qwen3.6 35B-A3B: 51.4 (#63)

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| GPQA Diamond | 56.1% | 84.8% |
| Confabulations | 29.4% | — |
| Vectara Hallucination Rate | 3.7% | — |
| LMArena Expert | 1203 | — |
| MMLU | 84.8% | — |

## Multilingual

- Phi-4: 37.4 (#232)
- Qwen3.6 35B-A3B: —

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| LMArena Non-English | 1197 | — |
| LMArena Chinese | 1211 | — |
| LMArena French | 1223 | — |
| LMArena German | 1222 | — |
| LMArena Japanese | 1158 | — |
| LMArena Korean | 1151 | — |
| LMArena Russian | 1209 | — |
| LMArena Spanish | 1234 | — |

## Instruction Following

- Phi-4: 60.6 (#246)
- Qwen3.6 35B-A3B: —

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| LiveBench Instruction Following | 58.4% | — |
| LMArena Instruction Following | 1201 | — |

## Long Context

- Phi-4: 36.9 (#221)
- Qwen3.6 35B-A3B: —

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| LMArena Longer Query | 1217 | — |

## Writing & Preference

- Phi-4: 40.7 (#239)
- Qwen3.6 35B-A3B: —

| Benchmark | Phi-4 | Qwen3.6 35B-A3B |
|---|---|---|
| LMArena Text | 1217 | — |
| LMArena Creative Writing | 1182 | — |
| Short-Story Creative Writing | 62.6% | — |
| LMArena Multi-Turn | 1206 | — |
| LiveBench Language | 25.6% | — |

## FAQ

### Is Phi-4 better than Qwen3.6 35B-A3B?

Qwen3.6 35B-A3B is the stronger model overall, scoring 37.2 to 30.9 on the Noometry Index. Phi-4 costs 6.4× less per token, which makes it the better buy when Qwen3.6 35B-A3B's lead doesn't matter for your workload.

### Which is cheaper, Phi-4 or Qwen3.6 35B-A3B?

Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Qwen3.6 35B-A3B lists at $0.25 and $1.49.

### Is Phi-4 or Qwen3.6 35B-A3B better for coding?

Qwen3.6 35B-A3B scores higher on coding benchmarks: 36.7 versus 33.8 in the Noometry coding category.

### Which has the bigger context window?

Qwen3.6 35B-A3B does, with 262K tokens against 128K.

### How many benchmarks do Phi-4 and Qwen3.6 35B-A3B share?

4 benchmarks have published results for both models. Phi-4 has 37 scored results on Noometry and Qwen3.6 35B-A3B has 14.