Model comparison

Ministral 8B vs Phi 3 Mini 4k Instruct

Ministral 8B and Phi 3 Mini 4k Instruct score almost the same on the Noometry Index (28.2 vs 27.9), so choose on price, context window or the category you care about most.

Last verified . 12 shared benchmarks.

Ministral 8B Mistral AI

28.2

Rank #325 Confirmed

Phi 3 Mini 4k Instruct Microsoft

27.9

Rank #328 Confirmed

Summary

  • They share 12 benchmarks with published results for both. Ministral 8B scores higher in 6 categories and Phi 3 Mini 4k Instruct in 2 categories; 7 gaps are clear of the uncertainty.
  • The widest gap is in knowledge, where Phi 3 Mini 4k Instruct leads 28.5 to 12.6.

Side by side

Ministral 8B and Phi 3 Mini 4k Instruct specifications
Ministral 8BPhi 3 Mini 4k Instruct
ProviderMistral AIMicrosoft
Noometry Index28.227.9
Released2024-10-012024-04-23
WeightsOpenOpen
Context window262K—
Max output262K—
Input $ / M tokens$0.15—
Output $ / M tokens$0.15—
Results tracked1735

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Ministral 8B leads

Ministral 8B: 35.0 (#230), Phi 3 Mini 4k Instruct: 26.6 (#323)

Coding benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
LMArena Coding12021093
LiveBench Coding—15.5%
HumanEval+—59.1%
MBPP+—54.2%

Agentic & Tool Use Not comparable

Ministral 8B: 16.4 (#148), Phi 3 Mini 4k Instruct: —

Agentic & Tool Use benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
Berkeley Function Calling Leaderboard11.1%—

Reasoning Ministral 8B leads

Ministral 8B: 18.4 (#281), Phi 3 Mini 4k Instruct: 14.1 (#328)

Reasoning benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
LMArena Hard Prompts11911072
Chess Puzzles—0%
LiveBench Reasoning—26.8%
DTBench45.7%—
LiveBench Data Analysis—34.7%
Adversarial NLI—52.8%
BIG-Bench Hard—71.7%
HellaSwag—76.7%
LiveBench—22.4%
WinoGrande—70.8%

Math Too close to call

Ministral 8B: 25.7 (#267), Phi 3 Mini 4k Instruct: 26.6 (#257)

Math benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
LMArena Math11881111
LiveBench Math—15.7%
MATH Level 514.9%—

Knowledge Phi 3 Mini 4k Instruct leads

Ministral 8B: 12.6 (#297), Phi 3 Mini 4k Instruct: 28.5 (#246)

Knowledge benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
LMArena Expert11701045
GPQA Diamond27.1%—
Vectara Hallucination Rate7.4%—
ARC (AI2) Challenge—84.9%
MMLU—68.8%
OpenBookQA—88%
TriviaQA—64%

Multilingual Ministral 8B leads

Ministral 8B: 35.1 (#247), Phi 3 Mini 4k Instruct: 26.3 (#280)

Multilingual benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
LMArena Non-English11651021
LMArena Chinese11931021
LMArena Russian11951022
LMArena French—1076
LMArena German—1044
LMArena Japanese—935
LMArena Korean—905
LMArena Spanish—1085

Instruction Following Ministral 8B leads

Ministral 8B: 60.5 (#250), Phi 3 Mini 4k Instruct: 47.7 (#303)

Instruction Following benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
LMArena Instruction Following11611053
LiveBench Instruction Following—39.1%

Long Context Ministral 8B leads

Ministral 8B: 36.7 (#227), Phi 3 Mini 4k Instruct: 31.7 (#276)

Long Context benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
LMArena Longer Query12121044

Writing & Preference Ministral 8B leads

Ministral 8B: 39.6 (#246), Phi 3 Mini 4k Instruct: 27.6 (#300)

Writing & Preference benchmarks
BenchmarkMinistral 8BPhi 3 Mini 4k Instruct
LMArena Text11911073
LMArena Creative Writing11751037
LMArena Multi-Turn11661018
LiveBench Language—9.2%

Frequently asked questions

Is Ministral 8B better than Phi 3 Mini 4k Instruct?

Ministral 8B and Phi 3 Mini 4k Instruct score almost the same on the Noometry Index (28.2 vs 27.9), so choose on price, context window or the category you care about most.

Is Ministral 8B or Phi 3 Mini 4k Instruct better for coding?

Ministral 8B scores higher on coding benchmarks: 35.0 versus 26.6 in the Noometry coding category.

How many benchmarks do Ministral 8B and Phi 3 Mini 4k Instruct share?

12 benchmarks have published results for both models. Ministral 8B has 17 scored results on Noometry and Phi 3 Mini 4k Instruct has 35.

Related comparisons

Go deeper