Compare · ModelsLive · 2 picked · head to head
Phi 2 vs MPT-30B
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Phi 2 wins on 10/13 benchmarks
Phi 2 wins 10 of 13 shared benchmarks. Leads in knowledge · reasoning · general.
Category leads
knowledge·Phi 2reasoning·Phi 2general·Phi 2language·Phi 2math·Phi 2
Hype vs Reality
Attention vs performance
Phi 2
#218 by perf·no signal
MPT-30B
#215 by perf·no signal
Vendor risk
Who is behind the model
Microsoft
$3.00T·Big Tech
U
Unknown
private · undisclosed
Head to head
13 benchmarks · 2 models
Phi 2MPT-30B
ARC AI2
Phi 2 leads by +33.7
AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval.
Phi 2
67.9
MPT-30B
34.1
BBH
Phi 2 leads by +28.5
BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans.
Phi 2
45.9
MPT-30B
17.3
HellaSwag
MPT-30B leads by +30.4
HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios.
Phi 2
38.1
MPT-30B
68.5
BBH (HuggingFace)
Phi 2 leads by +21.5
Phi 2
28.0
MPT-30B
6.5
GPQA
Phi 2 leads by +1.6
Phi 2
2.9
MPT-30B
1.3
IFEval
Phi 2 leads by +5.9
Phi 2
27.4
MPT-30B
21.5
MATH Level 5
Phi 2 leads by +1.4
Phi 2
3.0
MPT-30B
1.6
MMLU-PRO
Phi 2 leads by +15.8
Phi 2
18.1
MPT-30B
2.3
MUSR
Phi 2 leads by +10.9
Phi 2
13.8
MPT-30B
2.9
MMLU
Phi 2 leads by +14.0
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
Phi 2
44.5
MPT-30B
30.5
OpenBookQA
Phi 2 leads by +28.8
OpenBookQA · science questions that require combining a given core fact with broad common knowledge, mimicking an open-book exam setting.
Phi 2
64.8
MPT-30B
36.0
TriviaQA
MPT-30B leads by +28.4
TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents.
Phi 2
45.2
MPT-30B
73.6
Winogrande
MPT-30B leads by +32.6
WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs.
Phi 2
9.4
MPT-30B
42.0
Full benchmark table
| Benchmark | Phi 2 | MPT-30B |
|---|---|---|
ARC AI2 AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval. | 67.9 | 34.1 |
BBH BIG-Bench Hard · a curated subset of 23 challenging tasks from BIG-Bench where language models previously failed to outperform average humans. | 45.9 | 17.3 |
HellaSwag HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios. | 38.1 | 68.5 |
BBH (HuggingFace) | 28.0 | 6.5 |
GPQA | 2.9 | 1.3 |
IFEval | 27.4 | 21.5 |
MATH Level 5 | 3.0 | 1.6 |
MMLU-PRO | 18.1 | 2.3 |
MUSR | 13.8 | 2.9 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 44.5 | 30.5 |
OpenBookQA OpenBookQA · science questions that require combining a given core fact with broad common knowledge, mimicking an open-book exam setting. | 64.8 | 36.0 |
TriviaQA TriviaQA · reading comprehension benchmark with trivia questions, requiring models to find and reason over evidence from provided documents. | 45.2 | 73.6 |
Winogrande WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs. | 9.4 | 42.0 |