Model comparison

Mixtral 8x7B vs StarCoder 2 3B

Mixtral 8x7B has enough public results to be ranked (#334); StarCoder 2 3B does not yet, so treat this comparison as directional.

Last verified . 6 shared benchmarks.

Mixtral 8x7B Mistral AI

27.1

Rank #334 Confirmed

StarCoder 2 3B NVIDIA

36.3

Unranked Sparse

Summary

  • They share 6 benchmarks with published results for both. Mixtral 8x7B scores higher in 0 categories and StarCoder 2 3B in 1 category, but none of those gaps is larger than the uncertainty.

Side by side

Mixtral 8x7B and StarCoder 2 3B specifications
Mixtral 8x7BStarCoder 2 3B
ProviderMistral AINVIDIA
Noometry Index27.136.3
Released2023-12-112024-02-22
WeightsOpenOpen
Context window32K—
Max output32K—
Input $ / M tokens$0.70—
Output $ / M tokens$0.70—
Results tracked387

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

Category by category

Coding Too close to call

Mixtral 8x7B: 32.8 (#269), StarCoder 2 3B: 33.5

Coding benchmarks
BenchmarkMixtral 8x7BStarCoder 2 3B
HumanEval+39.6%27.4%
LMArena Coding1126—
BigCodeBench Complete—21.4%
MBPP+49.7%—

Reasoning Not comparable

Mixtral 8x7B: 18.2 (#285), StarCoder 2 3B: —

Reasoning benchmarks
BenchmarkMixtral 8x7BStarCoder 2 3B
Epoch Capabilities Index118.4788.76
WinoGrande77.2%57.1%
LMArena Hard Prompts1115—
DTBench49.6%—
Adversarial NLI55.2%—
ForecastBench56.3—
HellaSwag86.7%—
PIQA83.6%—

Math Not comparable

Mixtral 8x7B: 18.8 (#289), StarCoder 2 3B: —

Math benchmarks
BenchmarkMixtral 8x7BStarCoder 2 3B
GSM8K74.4%21.6%
Omni-MATH10.5%—
LMArena Math1147—
MATH Level 510%—

Knowledge Not comparable

Mixtral 8x7B: 11.0 (#301), StarCoder 2 3B: —

Knowledge benchmarks
BenchmarkMixtral 8x7BStarCoder 2 3B
ARC (AI2) Challenge87.3%34.2%
MMLU70.6%36.6%
GPQA Diamond30.6%—
MMLU-Pro33.5%—
GPQA (HELM)29.6%—
LMArena Expert1088—
OpenBookQA85.8%—
TriviaQA82.2%—

Multilingual Not comparable

Mixtral 8x7B: 29.6 (#266), StarCoder 2 3B: —

Multilingual benchmarks
BenchmarkMixtral 8x7BStarCoder 2 3B
LMArena Non-English1077—
LMArena Chinese1055—
LMArena French1166—
LMArena German1114—
LMArena Japanese931—
LMArena Korean968—
LMArena Russian1090—
LMArena Spanish1111—

Instruction Following Not comparable

Mixtral 8x7B: 51.0 (#297), StarCoder 2 3B: —

Instruction Following benchmarks
BenchmarkMixtral 8x7BStarCoder 2 3B
IFEval57.5%—
LMArena Instruction Following1109—

Long Context Not comparable

Mixtral 8x7B: 33.4 (#260), StarCoder 2 3B: —

Long Context benchmarks
BenchmarkMixtral 8x7BStarCoder 2 3B
LMArena Longer Query1103—

Writing & Preference Not comparable

Mixtral 8x7B: 34.2 (#270), StarCoder 2 3B: —

Writing & Preference benchmarks
BenchmarkMixtral 8x7BStarCoder 2 3B
LMArena Text1132—
LMArena Creative Writing1109—
WildBench67.3%—
LMArena Multi-Turn1115—

Frequently asked questions

Is Mixtral 8x7B better than StarCoder 2 3B?

Mixtral 8x7B has enough public results to be ranked (#334); StarCoder 2 3B does not yet, so treat this comparison as directional.

Is Mixtral 8x7B or StarCoder 2 3B better for coding?

They score almost the same on coding (32.8 vs 33.5); test both on your own repository before choosing.

How many benchmarks do Mixtral 8x7B and StarCoder 2 3B share?

6 benchmarks have published results for both models. Mixtral 8x7B has 38 scored results on Noometry and StarCoder 2 3B has 7.

Related comparisons

Go deeper