AFS (AI Foundry Service)

From training to deployment, the enterprise-grade large language model one-stop solution.
The world's first on-premises large language model solution, offering generative AI services with the highest cybersecurity standards.

AFS (AI Foundry Service)

Taiwan AI Cloud combines AIHPC computing power and a No-code large model optimization platform to launch a comprehensive enterprise-level generative AI solution and provides AFS series services that allow quick access to AI supercomputer computing power, easy model optimization, and flexible model deployment. These services include "AFS Platform," "the world's first AFS Appliance," and "AFS Cloud", and "AFS ModelSpace".

Taiwan AI Cloud provides reliable, energy-efficient, and easily migratable optimized AI models to allow enterprises to seamlessly move or adjust AI workloads between public and private clouds, with a low threshold and cost of software engineering, offering investment benefits that are tens of times higher compared to renting cloud services and thousands of times higher compared to building self-owned AIHPC systems. It helps enterprises quickly embrace the AI 2.0 application trend and put it into practice.

AFS (AI Foundry Service) Pipeline:One-stop Solution

AFS ModelSpace

Native Large Language Model Cloud Deployment Solution

Provide enterprises with a variety of open source large language models and flexibility of allocating computing resources.

✓ Easy to operate: Quickly deploy and start mainstream large language model AI services in the cloud.

✓ Cost transparency: Pay-as-you-go pricing. Simplize the pricing method to evaluate and control costs easier.

✓ Cloud hosting: Automation tools and management services that simplify the administration tasks of large language models.

ItemFoundation ModelsType
Native ModelFFM-MistralAFS.MS.FFM-Mixtral-8x7B
Native ModelFFM-Llama2-v2AFS.MS.FFM-Llama2-v2-70B
Native ModelFFM-Llama2AFS.MS.FFM-Llama2-70B
Native ModelFFM-BLOOMZAFS.MS.FFM-BLOOMZ-176B
Native ModelFFM-embeddingAFS.MS.FFM-embedding
Native ModelMeta-Llama2AFS.MS.Meta-Llama2-70B
Native ModelMeta-CodeLlamaAFS.MS.Meta-CodeLlama-34B-8K

*Please refer to the Pricing for complete products and prices.

AFS Platform

Large Language Model Optimization Service

Provide enterprises with the “lowest cost” and “highest efficiency” large language model training environment.

✓ Time and Cost Savings: Use a pre-trained FFM model to skip starting from scratch. Enterprises only need to prepare data to complete their LLM training quickly.

✓ Lower Entry Barriers: No need to build your own AI supercomputer ; pay as you go, and you can start large language model training tasks with costs ranging from only a few thousand to tens of thousands of yuan.

✓ Efficient Parameter Fine-tuning: Low-Rank Adaptation (LoRA) is available which reduces the memory usage and parameters required for training, balancing model performance with efficiency.

Fine-tuning StrategyFoundation ModelHighest specification product typePerformance
(Million Tokens/ hour)
Full parameterFFM-Llama2-v2-70BAFS.plt.FFM-Llama2-v2-70B.12M12
Full parameterFFM-Llama2-70BAFS.plt.FFM-Llama2-70B.12M12
Full parameterFFM-BLOOMZ-176BAFS.plt.FFM-BLOOMZ-176B.45M45
Full parameterMeta-Llama2-70BAFS.plt.Meta-Llama2-70B.12M12
LoRA (Low-Rank Adaptation)FFM-Llama2-v2-70BAFS.plt.FFM-Llama2-v2-LoRA-70B.16M16
LoRA (Low-Rank Adaptation)FFM-Llama2-70BAFS.plt.FFM-Llama2-LoRA-70B.16M16
LoRA (Low-Rank Adaptation)Meta-Llama2-70BAFS.plt.Meta-Llama2-LoRA-70B.16M16

*Please refer to the Pricing for complete products and prices.

〝Enterprise Brain Computing Plan〞
Specially designed for those needing massive computing power !
The longer the contract, the greater the savings.

PlanContract durationPrice
(NTD, Tax Exclusive)
Annual payment
(NTD, Tax Exclusive)
Total usable computing power
(NTD, Tax Exclusive)
Enterprise Brain 95One year950,000

950,0001,090,000
*Save NT$140,000*
Enterprise Brain 95Three year2,850,000 950,000
3,420,000
*Save NT$570,000*
Enterprise Brain 150One year1,500,000 1,500,000 1,800,000
*Save NT$300,000*
Enterprise Brain 150Three year4,500,000 1,500,000 5,700,000
*Save NT$1,200,000*
Enterprise Brain 250One year2,500,000 2,500,000 3,300,000
*Save NT$800,000*
Enterprise Brain 250Three year7,500,000 2,500,000 10,500,000
*Save NT$3,000,000*

*Please refer to the Pricing for complete products and prices.

FFM 176B vs. FFM 7B

FFM 176B vs. FFM 7B

AFS Cloud

Optimized Large Language Model Cloud Deployment Service

Offer flexible deployment solutions for enterprises, pay as you go to eliminate unnecessary costs.

✓ Simple and fast: Optimized large language model AI services can be quickly deployed in the cloud.

✓ Save time and effort: Cloud hosting allows you to immediately test the optimized model or put it into a production environment, without the hassle of data center management or significant setup costs.

✓ Affordable: Pay-as-you-go pricing, integrated deployment of optimized inference software, and easy cost control.

ItemFoundation ModelsType
Optimized ModelFFM-Llama2-v2AFS.cld.FFM-Llama2-v2-70B.ft
Optimized ModelFFM-Llama2AFS.cld.FFM-Llama2-70B.ft
Optimized ModelFFM-BLOOMZAFS.cld.FFM-BLOOMZ-176B.ft
Optimized ModelMeta-Llama2AFS.cld.Meta-Llama2-70B.ft

*Please refer to the Pricing for complete products and prices.

企業級大模型優化方案

AFS Shuttle

大型語言模型優化共乘服務

低成本快速投入,概念性驗證 (Proof of Concept,POC) 的最佳選擇

產品功能特色

✓ 容易上手的 No-code Portal

✓ 多種預訓練模型可供選擇

根據客戶需求和可讀取 token 總量排定製程

✓ 國家級的資安防護與租戶環境隔離

能進一步與安全線路環境整合(含 VPN )

✓ 提供包套固定價格給企業用戶快速取得 GPU 資源進行 POC 驗證

 

AFS Appliance

Large Language Model On-premises Solution

Provide enterprises with the ability to directly deploy their own large language models in local data centers/private clouds while ensuring both cybersecurity and privacy.

✓ Secure data storage: Compliant with enterprise on-premises security, compliance, and availability requirements, both input and output occur within the organization’s inference environment to ensure user and data security and privacy while minimizing cybersecurity risks.

✓ Confidential information stays secure: Enterprises can directly deploy their proprietary large models on their servers, without sharing environments with other users. This ensures that sensitive data remains protected and provides a high level of privacy and security.

✓ Flexible plans with special discounts: Various plan options to meet different requirements, with the highest-end option (DGX H100) providing a text generation performance of up to 34,000 tokens per minute.

TypeMain GPU spec. and config.ServerPrice
(NTD, Tax Exclusive)
dgxh8x NVIDIA H100 SMXNVIDIA DGX H100Contact us for more information.
8L40s8x NVIDIA L40s PCIeASUS ESC8000A Dual CPU3,970,000
4L40s4x NVIDIA L40s PCIeASUS ESC4000A Single CPU2,030,000
2L40s2x NVIDIA L40s PCIeASUS ESC4000A Single CPU1,300,000

*Please refer to the Pricing for complete products and prices.

〝AFS Appliance Joint Purchase Plan〞
You'll get UP to 2.04 MILLION DISCOUNTS & MORE services below !
* Free-of-charge installation and activation of the AFS Appliance
* Quarterly software and model updates
* 5 x 8 on-site hardware maintenance services

Appliance TypeEnterprise Brain 95
1 year contract
950,000
(NTD, Tax Exclusive)
Enterprise Brain 95
3 year contract
2,850,000
(NTD, Tax Exclusive)
Enterprise Brain 150
1 year contract
1,500,000
(NTD, Tax Exclusive)
Enterprise Brain 150
3 year contract
4,500,000
(NTD, Tax Exclusive)
Enterprise Brain 250
1 year contract
2,500,000
(NTD, Tax Exclusive)
Enterprise Brain 250
3 year contract
7,500,000
(NTD, Tax Exclusive)
dgxhJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discount
8L40sJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discount
4L40sJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discount
2L40sJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discountJoint purchase discount

*Please refer to the Pricing for complete products and prices.

「AFS (AI Foundry Service)」提供一站式整合服務,協助企業開發專屬的企業級生成式AI大模型

立即填表諮詢

Free Consultation Service

Contact our experts to learn more and get started with the best solution for you.

Apply for AFS

AFS (AI Foundry Service)
方案諮詢

Business Inquiries/ Distribution Cooperation