vs
3:9
overall capability wins (1 ties across 13 scored). Phi-4 leads overall.
- Reasoning2:4
- Coding0:3
- Arena0:1
- Pricing0:3
- Speed0:2
- Specs1:1
Verdict
Phi-4 leads coding (widest gap: +17.0 on SciCode); Phi-4 ranks higher on LMArena (+128 Elo); Phi-3 Mini 3.8B offers ~7.8x larger context (128K vs 16K).
- Phi-4 leads coding (widest gap: +17.0 on SciCode)
- Phi-4 ranks higher on LMArena (+128 Elo)
- Phi-3 Mini 3.8B offers ~7.8x larger context (128K vs 16K)
- Phi-4 is ~2.6x cheaper on a blended token basis than Phi-3 Mini 3.8B
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Phi-3 Mini 3.8B
- $0.26
- Phi-4
- $0.11
Phi-4 is about 2.5x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Microsoft
Microsoft
Family
Phi
Phi
License
MIT
MIT
Open weights
Yes
Yes
Release
Apr 23, 2024
Dec 12, 2024
Knowledge cutoff
-
-
API / provider
Azure AI Foundry
Azure / self-host
Modalities
text → text
text → text
Specs
Context window
128K
16K
Max output
4K
4K
Parameters
3.8B
14B
Pricing
Input $/1M
$0.13
$0.07
Output $/1M
$0.52
$0.14
Blended $/1M (3∶1)
$0.23
$0.09
Speed
tok/s
140
150
TTFT (s)
0.18
0.15
Reasoning
MMLU-Pro
43.5
70.4
GPQA Diamond
31.9
57.5
Humanity's Last Exam
5.0
3.8
AIME 2025
0.3
18.0
MATH-500
45.7
81.0
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
—
14.1
AA-LCR v1.1
—
—
CritPt
—
—
MMMU-Pro
—
—
IFBench
23.9
23.5
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
11.6
23.1
Terminal-Bench 2.1
—
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
29.6
45.5
SciCode
9.0
26.0
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,128
1,256
| Metric | Phi-3 Mini 3.8B | Phi-4 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Microsoft | Microsoft | — |
| Family | Phi | Phi | — |
| License | MIT | MIT | — |
| Open weights | Yes | Yes | — |
| Release | Apr 23, 2024 | Dec 12, 2024 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Azure AI Foundry | Azure / self-host | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 128K | 16K | A +112K |
| Max output | 4K | 4K | tie |
| Parameters | 3.8B | 14B | B +10 |
| Pricing | |||
| Input $/1M | $0.13 | $0.07 | B +$0.06 |
| Output $/1M | $0.52 | $0.14 | B +$0.38 |
| Blended $/1M (3∶1) | $0.23 | $0.09 | B +$0.14 |
| Speed | |||
| tok/s | 140 | 150 | B +10 |
| TTFT (s) | 0.18 | 0.15 | B +0.03 |
| Reasoning | |||
| MMLU-Pro | 43.5 | 70.4 | B +26.9 pts |
| GPQA Diamond | 31.9 | 57.5 | B +25.6 pts |
| Humanity's Last Exam | 5.0 | 3.8 | A +1.2 pts |
| AIME 2025 | 0.3 | 18.0 | B +17.7 pts |
| MATH-500 | 45.7 | 81.0 | B +35.3 pts |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | — | 14.1 | — |
| AA-LCR v1.1 | — | — | — |
| CritPt | — | — | — |
| MMMU-Pro | — | — | — |
| IFBench | 23.9 | 23.5 | A +0.4 pts |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | 11.6 | 23.1 | B +11.5 pts |
| Terminal-Bench 2.1 | — | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | 29.6 | 45.5 | B +15.9 pts |
| SciCode | 9.0 | 26.0 | B +17.0 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,128 | 1,256 | B +128 Elo |
Phi-3 Mini 3.8B
- Context
- 128K / 4K out
- Parameters
- 3.8B
- Price
- $0.13 / $0.52
- Speed
- 140 tok/s · 0.18s TTFT
- Modalities
- text → text
- License
- MIT
Original Phi-3 Mini instruct — compact Azure / edge SLM that popularized Microsoft's open Phi line.
Phi-4
- Context
- 16K / 4K out
- Parameters
- 14B
- Price
- $0.07 / $0.14
- Speed
- 150 tok/s · 0.15s TTFT
- Modalities
- text → text
- License
- MIT
Small high-quality dense model from Microsoft - punches above its size on STEM.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.