Skip to content
Updated minutes ago
Google

Gemma 2 2B

Google · dense · 2.6B parameters · 8,192 context

Quality
44.0

Parameters

2.6B

Context Window

8K tokens

Architecture

Dense

Best GPU

RTX 3080

Quality Score

44/100

Intelligence Brief

Gemma 2 2B is a 2.6B parameter DENSE model from Google, featuring Grouped Query Attention (GQA) with 26 layers and 2,304 hidden dimensions. With a 8,192 token context window, it supports structured output, code, math, multilingual. On standardized benchmarks, it achieves MMLU 52.2, HumanEval 25, GSM8K 48. For self-hosted inference, RTX 3080 delivers optimal throughput at $133/month.

Architecture Details

TypeDENSE
Total Parameters2.6B
Active Parameters2.6B
Layers26
Hidden Dimension2,304
Attention Heads8
KV Heads4
Head Dimension256
Vocab Size256,000

Memory Requirements

BF16 Weights

5.2 GB

FP8 Weights

2.6 GB

INT4 Weights

1.3 GB

KV-Cache per Token106496 bytes
Activation Estimate0.30 GB

GPU Compatibility Matrix

Gemma 2 2B is compatible with 100% of GPU configurations across 41 GPUs at 3 precision levels.

BF16 (Full)
FP8 (Half)
INT4 (Quarter)
Blackwell(7 GPUs)
B200 NVL (pair)360GB
B300288GB
B100 SXM192GB
GB200 NVL72 (per GPU)192GB
Hopper(7 GPUs)
H100 NVL 94GB (per GPU pair)188GB
H200 SXM141GB
H2096GB
GH20096GB
Ada Lovelace(11 GPUs)
L40S48GB
L4048GB
RTX 6000 Ada48GB
L2048GB
Ampere(16 GPUs)
A100 80GB SXM80GB
A100 80GB PCIe80GB
A1664GB
RTX A600048GB
Legend:No fitVery tightTightModerateGoodExcellent

GPU Recommendations

RTX 3080optimal

BF16 · 1 GPU · vllm

100/100

score

Throughput

710.3 tok/s

Latency (ITL)

1.4ms

Est. TTFT

0ms

Cost/Month

$133

Cost/M Tokens

$0.07

Use this config →
RTX 4060optimal

BF16 · 1 GPU · vllm

100/100

score

Throughput

254.2 tok/s

Latency (ITL)

3.9ms

Est. TTFT

1ms

Cost/Month

$209

Cost/M Tokens

$0.31

Use this config →
RTX 3070optimal

BF16 · 1 GPU · vllm

100/100

score

Throughput

418.7 tok/s

Latency (ITL)

2.4ms

Est. TTFT

0ms

Cost/Month

$85

Cost/M Tokens

$0.08

Use this config →

Deployment Options

API

API Deployment

No API pricing available

Self-Hosted

Single GPU

RTX 3080

$133/mo

Min VRAM: 3 GB

Scale

Multi-GPU

RTX 3080

710.3 tok/s

Best available config

API Pricing Comparison

No API pricing data available for this model.

Performance Estimates

Throughput by GPU

RTX 3080
710.3 tok/s
RTX 4060
254.2 tok/s
RTX 3070
418.7 tok/s

VRAM Breakdown (RTX 3080, BF16)

Weights
KV
Act
Weights 5.2 GBKV-Cache 1.7 GBActivations 2.4 GBOverhead 0.4 GB

Precision Impact

bf16

5.2 GB

weights/GPU

~710.3 tok/s

fp8

2.6 GB

weights/GPU

int4

1.3 GB

weights/GPU

Quality Benchmarks

Bottom 25%
5th percentile across all models
MMLU
52.2
Bottom 25% (12th pctile)
HumanEval
25.0
Bottom 25% (6th pctile)
GSM8K
48.0
Bottom 25% (15th pctile)
MT-Bench
65.0
Bottom 25% (0th pctile)

Capabilities

Features

Tool Use Vision Code Math Reasoning Multilingual Structured Output

Supported Frameworks

vllmsglangtgitensorrt-llmollama

Supported Precisions

BF16 (default)FP8INT4

Where to Deploy Gemma 2 2B

Similar Models

Frequently Asked Questions

How much VRAM does Gemma 2 2B need for inference?

Gemma 2 2B requires approximately 5.2 GB of VRAM at BF16 precision, 2.6 GB at FP8, or 1.3 GB at INT4 quantization. Additional VRAM is needed for KV-cache (106496 bytes per token) and activations (~0.30 GB).

What is the best GPU for Gemma 2 2B?

The top recommended GPU for Gemma 2 2B is the RTX 3080 using BF16 precision. It achieves approximately 710.3 tokens/sec at an estimated cost of $133/month ($0.07/M tokens). Score: 100/100.

How much does Gemma 2 2B inference cost?

Gemma 2 2B inference costs vary by provider and GPU setup. Use our calculator for detailed cost estimates across all providers.