Skip to content
Updated minutes ago
TII

Falcon 3 10B

TII UAE · dense · 10.3B parameters · 32,768 context

Quality
50.0

Falcon 3 10B is a 10.3B parameter DENSE model from TII UAE, featuring a 32,768 token context window. With 40 transformer layers and a hidden dimension of 4,096, it delivers efficient Grouped Query Attention (GQA) for optimized inference throughput. Based on InferenceBench analysis, the optimal deployment configuration is the A100 40GB SXM at BF16 precision, achieving approximately 407.6 tokens/second at $0.75/million tokens.

Architecture Details

TypeDENSE
Total Parameters10.3B
Active Parameters10.3B
Layers40
Hidden Dimension4,096
Attention Heads32
KV Heads8
Head Dimension128
Vocab Size131,072

Memory Requirements

BF16 Weights

20.6 GB

FP8 Weights

10.3 GB

INT4 Weights

5.2 GB

KV-Cache per Token81920 bytes
Activation Estimate0.70 GB

Fits on (single-node)

B200 SXM BF16B100 SXM BF16GB200 NVL72 (per GPU) BF16GB300 NVL72 (per GPU) BF16H200 SXM BF16H100 SXM BF16H100 PCIe BF16H100 NVL BF16

GPU Recommendations

A100 40GB SXMoptimal

BF16 · 1 GPU · vllm

95/100

score

Throughput

407.6 tok/s

Cost/Month

$807

Cost/M Tokens

$0.75

Use this config →
RTX 5090optimal

BF16 · 1 GPU · vllm

95/100

score

Throughput

469.7 tok/s

Cost/Month

$845

Cost/M Tokens

$0.68

Use this config →
A100 40GB PCIeoptimal

BF16 · 1 GPU · vllm

95/100

score

Throughput

407.6 tok/s

Cost/Month

$655

Cost/M Tokens

$0.61

Use this config →

API Pricing Comparison

No API pricing data available for this model.

Capabilities

Features

Tool Use Vision Code Math Reasoning Multilingual Structured Output

Supported Frameworks

vllmsglangtgillama-cppollama

Supported Precisions

BF16 (default)FP8INT4

Similar Models

Frequently Asked Questions

How much VRAM does Falcon 3 10B need for inference?

Falcon 3 10B requires approximately 20.6 GB of VRAM at BF16 precision, 10.3 GB at FP8, or 5.2 GB at INT4 quantization. Additional VRAM is needed for KV-cache (81920 bytes per token) and activations (~0.70 GB).

What is the best GPU for Falcon 3 10B?

The top recommended GPU for Falcon 3 10B is the A100 40GB SXM using BF16 precision. It achieves approximately 407.6 tokens/sec at an estimated cost of $807/month ($0.75/M tokens). Score: 95/100.

How much does Falcon 3 10B inference cost?

Falcon 3 10B inference costs vary by provider and GPU setup. Use our calculator for detailed cost estimates across all providers.