The AI arena is free today

Open Superagent

Phi 4 Mini vs Phi 4 Reasoning Plus

Phi 4 Reasoning Plus leads the LLM Stats Score 15.0 to -4.1.

Microsoft · Microsoft · Updated for 2026

Which is better?

Phi 4 Reasoning Plus leads the overall LLM Stats Score 15.0 to -4.1, ranking #247 overall.

In the 3 individual benchmarks reported for both models, Phi 4 Reasoning Plus wins 3; this is a narrower head-to-head signal than the composite indexes.

Based on current LLM Stats indexes, shared benchmarks, pricing, and model metadata for 2026.

Choose Phi 4 Mini

  • you are already invested in the Microsoft ecosystem

Choose Phi 4 Reasoning Plus

  • overall performance matters — it scores 15.0 and ranks #247 on LLM Stats
  • your work emphasizes reasoning — it leads those capability indexes
  • you value its reported benchmark strengths — it wins 3 of 3 exact shared results
  • you want the most recent training data — it shipped Apr 2025

At a glance

The differences that matter most.

Core performance indexes
-4.1
#363
15.0
#247
-4.2
#356
15.4
#240
Cost, coverage & limits
Benchmark wins
0 of 3
3 of 3
Input price
— / M
— / M
Output price
— / M
— / M
Context window
—
—

Capability indexes

Additional strengths measured across groups of related public benchmarks

1 shared
Index
Phi 4 Mini
Phi 4 Reasoning Plus
2.3#299
18.3#188
Conservative TrueSkill rating · higher is betterHow scores work

Individual benchmarks

17 reported for Phi 4 Mini · 11 for Phi 4 Reasoning Plus

3 shared

Phi 4 Mini outperforms in 0 benchmarks, while Phi 4 Reasoning Plus is better at 3 benchmarks (Arena Hard, GPQA, MMLU-Pro).

Phi 4 Reasoning Plus significantly outperforms across most benchmarks.

Sat Oct 10 2026 • llm-stats.com

Human preference

Blind head-to-head votes and playground preference scores

Model Size

Parameter count comparison

10.2B diff

Phi 4 Reasoning Plus has 10.2B more parameters than Phi 4 Mini, making it 264.6% larger.

Microsoft
Phi 4 Mini
3.8Bparameters
Microsoft
Phi 4 Reasoning Plus
14.0Bparameters
3.8B
Phi 4 Mini
14.0B
Phi 4 Reasoning Plus

License

Usage and distribution terms

Both models are licensed under MIT.

Both models share the same licensing terms, providing consistent usage rights.

Phi 4 Mini

MIT

Open weights

Phi 4 Reasoning Plus

MIT

Open weights

Release Timeline

When each model was launched

Phi 4 Mini was released on 2025-02-01, while Phi 4 Reasoning Plus was released on 2025-04-30.

Phi 4 Reasoning Plus is 3 months newer than Phi 4 Mini.

Phi 4 Mini

Feb 1, 2025

1.7 years ago

Phi 4 Reasoning Plus

Apr 30, 2025

1.4 years ago

2mo newer

Knowledge Cutoff

When training data ends

Phi 4 Mini has a knowledge cutoff of 2024-06-01, while Phi 4 Reasoning Plus has a cutoff of 2025-03-01.

Phi 4 Reasoning Plus has more recent training data (up to 2025-03-01), making it potentially better informed about events through that date compared to Phi 4 Mini (2024-06-01).

Phi 4 Mini

Jun 2024

Phi 4 Reasoning Plus

Mar 2025

9 mo newer

Outputs Comparison

Notice missing or incorrect data?

Judge for yourself.

Run your own prompts against Phi 4 Mini and Phi 4 Reasoning Plus side-by-side, then vote on the output you prefer.

Phi 4 Mini
✓ Preferred
Phi 4 Reasoning Plus
Open in Playground

FAQ

Common questions about Phi 4 Mini vs Phi 4 Reasoning Plus.

Which is better, Phi 4 Mini or Phi 4 Reasoning Plus?

Phi 4 Reasoning Plus leads the LLM Stats Score 15.0 to -4.1. Phi 4 Mini is made by Microsoft and Phi 4 Reasoning Plus is made by Microsoft. The best choice depends on your use case — compare their capability indexes, individual benchmarks, pricing, and limits above.

How does Phi 4 Mini compare to Phi 4 Reasoning Plus in benchmarks?

Phi 4 Mini scores GSM8k: 88.6%, ARC-C: 83.7%, BoolQ: 81.2%, OpenBookQA: 79.2%, PIQA: 77.6%. Phi 4 Reasoning Plus scores FlenQA: 97.9%, HumanEval+: 92.3%, IFEval: 84.9%, OmniMath: 81.9%, AIME 2024: 81.3%.

What are the main differences between Phi 4 Mini and Phi 4 Reasoning Plus?

Key differences include LLM Stats Score (-4.1 vs 15.0). See the full comparison above for benchmark-by-benchmark results.