Compare · ModelsLive · 2 picked · head to head
Phi 4 Mini Instruct vs Phi 4
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Phi 4 wins on 9/16 benchmarks
Phi 4 wins 9 of 16 shared benchmarks. Leads in general · knowledge · math.
Category leads
speed·Phi 4 Mini Instructgeneral·Phi 4knowledge·Phi 4language·Phi 4 Mini Instructmath·Phi 4reasoning·Phi 4
Hype vs Reality
Attention vs performance
Phi 4 Mini Instruct
#258 by perf·no signal
Phi 4
#202 by perf·#20 by attention
Vendor risk
Who is behind the model
Microsoft
$3.96T·Big Tech
Microsoft
$3.96T·Big Tech
Head to head
16 benchmarks · 2 models
Phi 4 Mini InstructPhi 4
Artificial Analysis · Agentic Index
Phi 4 Mini Instruct leads by +2.7
Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?"
Phi 4 Mini Instruct
2.7
Phi 4
0.0
Artificial Analysis · Coding Index
Phi 4 leads by +7.6
Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads.
Phi 4 Mini Instruct
3.6
Phi 4
11.2
Artificial Analysis · CritPt
Phi 4 Mini Instruct
0.0
Phi 4
0.0
Artificial Analysis · GPQA Diamond
Phi 4 leads by +24.4
Phi 4 Mini Instruct
33.1
Phi 4
57.5
Artificial Analysis · Humanity's Last Exam
Phi 4 Mini Instruct leads by +0.7
Phi 4 Mini Instruct
4.5
Phi 4
3.8
Artificial Analysis · IFBench
Phi 4 leads by +2.4
Phi 4 Mini Instruct
21.1
Phi 4
23.5
Artificial Analysis · Long Context Reasoning
Phi 4 Mini Instruct leads by +15.3
Phi 4 Mini Instruct
15.3
Phi 4
0.0
Artificial Analysis · Quality Index
Phi 4 Mini Instruct leads by +0.4
Phi 4 Mini Instruct
6.3
Phi 4
5.9
Artificial Analysis · tau2-Bench Telecom
Phi 4 Mini Instruct leads by +8.2
Phi 4 Mini Instruct
8.2
Phi 4
0.0
Artificial Analysis · Terminal-Bench Hard
Phi 4 leads by +3.8
Phi 4 Mini Instruct
0.0
Phi 4
3.8
BBH (HuggingFace)
Phi 4 leads by +16.9
Phi 4 Mini Instruct
38.7
Phi 4
55.7
GPQA
Phi 4 leads by +1.2
Phi 4 Mini Instruct
7.9
Phi 4
9.2
IFEval
Phi 4 Mini Instruct leads by +8.9
Phi 4 Mini Instruct
73.8
Phi 4
64.8
MATH Level 5
Phi 4 leads by +28.3
Phi 4 Mini Instruct
17.0
Phi 4
45.2
MMLU-PRO
Phi 4 leads by +15.8
Phi 4 Mini Instruct
32.6
Phi 4
48.3
MUSR
Phi 4 leads by +5.0
Phi 4 Mini Instruct
6.5
Phi 4
11.4
Full benchmark table
| Benchmark | Phi 4 Mini Instruct | Phi 4 |
|---|---|---|
Artificial Analysis · Agentic Index Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?" | 2.7 | 0.0 |
Artificial Analysis · Coding Index Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads. | 3.6 | 11.2 |
Artificial Analysis · CritPt | 0.0 | 0.0 |
Artificial Analysis · GPQA Diamond | 33.1 | 57.5 |
Artificial Analysis · Humanity's Last Exam | 4.5 | 3.8 |
Artificial Analysis · IFBench | 21.1 | 23.5 |
Artificial Analysis · Long Context Reasoning | 15.3 | 0.0 |
Artificial Analysis · Quality Index | 6.3 | 5.9 |
Artificial Analysis · tau2-Bench Telecom | 8.2 | 0.0 |
Artificial Analysis · Terminal-Bench Hard | 0.0 | 3.8 |
BBH (HuggingFace) | 38.7 | 55.7 |
GPQA | 7.9 | 9.2 |
IFEval | 73.8 | 64.8 |
MATH Level 5 | 17.0 | 45.2 |
MMLU-PRO | 32.6 | 48.3 |
MUSR | 6.5 | 11.4 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| — | — | — | — | |
| $0.07 | $0.14 | 16K tokens (~8 books) | $0.88 |