Skip to content

Mixtral 8x22B vs Gemma 3 27B

Mistral
Mixtral 8x22B

Mistral AI · 141B params · Quality: 73

Google
Gemma 3 27B

Google · 27B params · Quality: 76

Architecture Comparison

SpecMixtral 8x22BGemma 3 27B
TypeMOEDENSE
Total Parameters141B27B
Active Parameters39B27B
Layers5662
Hidden Dimension6,1443,584
Attention Heads4832
KV Heads816
Context Length65,536131,072
Precision (default)BF16BF16
Total Experts8N/A
Active Experts2N/A

Memory Requirements

PrecisionMixtral 8x22BGemma 3 27B
BF16 Weights282.0 GB54.0 GB
FP8 Weights141.0 GB27.0 GB
INT4 Weights70.5 GB13.5 GB
KV-Cache / Token229376 B507904 B
Activation Estimate2.50 GB1.50 GB

Minimum GPUs Needed (BF16)

H100 SXM5 GPUs1 GPU
L40S7 GPUs2 GPUs

Quality Benchmarks

BenchmarkMixtral 8x22BGemma 3 27B
Overall7376
MMLU77.878.0
HumanEval46.048.0
GSM8K78.485.0
MT-Bench80.082.0

Mixtral 8x22B

MMLU
77.8
HumanEval
46.0
GSM8K
78.4
MT-Bench
80.0

Gemma 3 27B

MMLU
78.0
HumanEval
48.0
GSM8K
85.0
MT-Bench
82.0

Capabilities

FeatureMixtral 8x22BGemma 3 27B
Tool Use✓ Yes✓ Yes
Vision✗ No✓ Yes
Code✓ Yes✓ Yes
Math✓ Yes✓ Yes
Reasoning✗ No✗ No
Multilingual✓ Yes✓ Yes
Structured Output✓ Yes✓ Yes

API Pricing Comparison

Cheapest Output (Mixtral 8x22B)

$1.20/M

Input: $1.20/M

Cheapest Output (Gemma 3 27B)

$0.20/M

Input: $0.10/M

ProviderMixtral 8x22B In $/MOut $/MGemma 3 27B In $/MOut $/M
google$0.10$0.20
together$1.20$1.20$0.30$0.30
mistral$2.00$6.00

Recommendation Summary

  • Gemma 3 27B scores higher on overall quality (76 vs 73).
  • Gemma 3 27B is cheaper per output token ($0.20/M vs $1.20/M).
  • Gemma 3 27B has a smaller memory footprint (54.0 GB vs 282.0 GB BF16), making it easier to deploy on fewer GPUs.
  • Gemma 3 27B supports a longer context window (131,072 vs 65,536 tokens).
  • Mixtral 8x22B uses MOE architecture while Gemma 3 27B uses DENSE. MoE models activate fewer parameters per token, improving inference efficiency.
  • Gemma 3 27B is stronger at code generation (HumanEval: 48.0 vs 46.0).
  • Gemma 3 27B is better at math reasoning (GSM8K: 85.0 vs 78.4).

Compare Other Models