Skip to content
Updated minutes ago
InternLM

InternVL2 26B

InternLM · dense · 26B parameters · 32,768 context

Quality
50.0

Parameters

26B

Context Window

32K tokens

Architecture

Dense

Best GPU

H20

Intelligence Brief

InternVL2 26B is a 26B parameter DENSE model from InternLM, featuring Grouped Query Attention (GQA) with 48 layers and 6,144 hidden dimensions. With a 32,768 token context window, it supports vision, code, math, multilingual. For self-hosted inference, H20 delivers optimal throughput at $940/month.

Architecture Details

TypeDENSE
Total Parameters26B
Active Parameters26B
Layers48
Hidden Dimension6,144
Attention Heads48
KV Heads8
Head Dimension128
Vocab Size92,544

Memory Requirements

BF16 Weights

52.0 GB

FP8 Weights

26.0 GB

INT4 Weights

13.0 GB

KV-Cache per Token196608 bytes
Activation Estimate1.50 GB

GPU Compatibility Matrix

InternVL2 26B is compatible with 62% 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

H20optimal

FP8 · 1 GPU · tensorrt-llm

100/100

score

Throughput

1.1K tok/s

Latency (ITL)

1.0ms

Est. TTFT

0ms

Cost/Month

$940

Cost/M Tokens

$0.34

Use this config →
H100 SXMoptimal

FP8 · 1 GPU · tensorrt-llm

95/100

score

Throughput

1.1K tok/s

Latency (ITL)

1.0ms

Est. TTFT

0ms

Cost/Month

$1794

Cost/M Tokens

$0.65

Use this config →
H100 PCIeoptimal

FP8 · 1 GPU · tensorrt-llm

95/100

score

Throughput

643.5 tok/s

Latency (ITL)

1.6ms

Est. TTFT

0ms

Cost/Month

$1794

Cost/M Tokens

$1.06

Use this config →

Deployment Options

API

API Deployment

No API pricing available

Self-Hosted

Single GPU

H20

$940/mo

Min VRAM: 26 GB

Scale

Multi-GPU

A100 40GB SXM x2

317.8 tok/s

TP· $1613/mo

API Pricing Comparison

No API pricing data available for this model.

Performance Estimates

Throughput by GPU

H20
1.1K tok/s
H100 SXM
1.1K tok/s
H100 PCIe
643.5 tok/s

VRAM Breakdown (H20, FP8)

Weights
Act
Weights 26.0 GBKV-Cache 1.6 GBActivations 12.0 GBOverhead 1.3 GB

Precision Impact

bf16

52.0 GB

weights/GPU

fp8

26.0 GB

weights/GPU

~1.1K tok/s

int4

13.0 GB

weights/GPU

Capabilities

Features

Tool Use Vision Code Math Reasoning Multilingual Structured Output

Supported Frameworks

vllmsglangtgi

Supported Precisions

BF16 (default)FP8INT4

Where to Deploy InternVL2 26B

Similar Models

Gemma 2 27B

27B params · dense

Quality: 65

from $0.27/M

Smaller context, Higher qualityCompare →

Gemma 3 27B

27B params · dense

Quality: 69

from $0.20/M

Larger context, Higher qualityCompare →
Higher qualityCompare →
Larger contextCompare →

Codestral 22B

22B params · dense

Quality: 63

from $0.90/M

Higher qualityCompare →

Frequently Asked Questions

How much VRAM does InternVL2 26B need for inference?

InternVL2 26B requires approximately 52.0 GB of VRAM at BF16 precision, 26.0 GB at FP8, or 13.0 GB at INT4 quantization. Additional VRAM is needed for KV-cache (196608 bytes per token) and activations (~1.50 GB).

What is the best GPU for InternVL2 26B?

The top recommended GPU for InternVL2 26B is the H20 using FP8 precision. It achieves approximately 1.1K tokens/sec at an estimated cost of $940/month ($0.34/M tokens). Score: 100/100.

How much does InternVL2 26B inference cost?

InternVL2 26B inference costs vary by provider and GPU setup. Use our calculator for detailed cost estimates across all providers.