Rent Turkey's most powerful cloud GPU infrastructure for your AI projects

Hourly and monthly dedicated GPU servers in our Istanbul Tier-IV datacenter for model training, fine-tuning, and large-scale inference.

Explore GPU Models → Explore GPU Pricing

Access all open-weight models with a single OpenAI-compatible API

Connect to 10+ models instantly without changing your code, running on our high-performance local infrastructure.

Browse model catalog → View plans

Change only the base_url and the API key

# curl
curl https://llm.fixcloud.ai/v1/chat/completions \
  -H "Authorization: Bearer $FXC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-oss-120b",
    "messages": [
      {"role": "user", "content": "Hello!"}
    ],
    "stream": true
  }'

# Python, official OpenAI SDK
from openai import OpenAI

client = OpenAI(
    base_url="https://llm.fixcloud.ai/v1",
    api_key="FXC_API_KEY",
)
r = client.chat.completions.create(
    model="gpt-oss-120b",
    messages=[{"role": "user",
               "content": "Hello!"}],
    stream=True,
)

// Node.js, official OpenAI SDK
import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://llm.fixcloud.ai/v1',
  apiKey: process.env.FXC_API_KEY,
});
const r = await client.chat.completions.create({
  model: 'gpt-oss-120b',
  messages: [{ role: 'user',
               content: 'Hello!' }],
  stream: true,
});

Run serverless vLLM inference with your own models

Zero idle costs when there is no traffic, sub-second cold starts on the first request, and maximum token throughput powered by vLLM.

Discover serverless → Get started now

FixCloud Serverless vLLM Inference Engine

High-Performance GPU Server Fleet

Cover your AI development, machine learning and high-performance computing needs without investing in hardware: NVIDIA GH200 Grace Hopper and RTX PRO 6000 Blackwell GPUs on HPE servers

Latest Hardware

Reach the newest accelerator technology with the NVIDIA GH200 Grace Hopper Superchip and RTX PRO 6000 Blackwell, and scale your AI application as far as you need.

Ready Right Away

Pick your GPU resource in the console.fixcloud.ai web interface and start working within minutes over secure access.

Cost Advantage

No upfront investment and no commitment: with hourly use you pay only for what you use, and a monthly commitment brings much lower prices.

Data Sovereignty

Your data stays within Turkey, on 100% KVKK-compliant FixCloud infrastructure in the Equinix IL4 data center in Istanbul; you reach your GPU resources through a layered security architecture.

Flagship

NVIDIA GH200 NVL2

288 GB HBM3e

Architecture 2 × Grace Hopper Superchip
Processor 144 Arm Neoverse V2 cores
Unified Memory 288 GB HBM3e + 960 GB LPDDR5X
NVLink-C2C (CPU–GPU) 900 GB/s per superchip
Server HPE ProLiant DL384 Gen12

On-Demand

$9.99 / hr

$6,999 / mo committed

Immediate Delivery Rent now →

Next-Gen Blackwell

RTX PRO 6000 Blackwell

96 GB GDDR7

Architecture NVIDIA Blackwell (Server Edition)
CUDA Cores 24,064
GPU Memory 96 GB GDDR7 (ECC)
Memory Bandwidth 1,597 GB/s
Server HPE ProLiant DL380a Gen12 · 1, 2 or 4 cards

On-Demand · 1 Card

$2.99 / hr

$1,999 / mo committed

Pre-Order Rent now →

GPUaaS Pricing & Specs

List, on-demand and committed prices for dedicated server, VM / container and serverless options

GPU Model	Cards	VRAM	Service Model	List (Hourly)	On-Demand (Hourly)	Committed (Monthly)	Action
NVIDIA GH200 Grace Hopper NVL2 2 × Superchip	2x GPU	288 GB	Dedicated HPE DL 384 G12	$19.99	$9.99	$6,999	Rent →
NVIDIA GH200 Grace Hopper 1 × Superchip	1x GPU	144 GB	Dedicated VM / Container	$9.99	$5.99	$3,999	Rent →
NVIDIA GH200 Grace Hopper 4 Core · 72 GB MIG	MIG	72 GB	Container / Serverless	$6.99	$3.99	$2,199	Rent →
NVIDIA GH200 Grace Hopper 2 Core · 36 GB MIG	MIG	36 GB	Container / Serverless	$2.49	$1.99	$1,499	Rent →
NVIDIA GH200 Grace Hopper 1 Core · 18 GB MIG	MIG	18 GB	Container / Serverless	$1.99	$1.49	$999	Rent →
NVIDIA RTX PRO 6000 Blackwell 1 card	1x GPU	96 GB	Dedicated HPE DL 380a	$3.49	$2.99	$1,999	Rent →
NVIDIA RTX PRO 6000 Blackwell 2 cards	2x GPU	192 GB	Dedicated HPE DL 380a	$9.49	$5.99	$4,999	Rent →
NVIDIA RTX PRO 6000 Blackwell 4 cards	4x GPU	384 GB	Dedicated HPE DL 380a	$21.49	$12.99	$7,999	Rent →

SUPERINTELLIGENCE CLOUD · TURKEY

LLMaaS: One API, All Models

Integrate large language models (LLMs) into your applications in minutes through a single API.

10+ Ready Models

256K Context

100% KVKK Compliant

Datasheet'i indir (PDF)

OpenAI

Qwen

deepseek

Mistral AI

What is LLMaaS?

FixCloud LLMaaS delivers 10+ ready foundation models—including GPT-OSS, DeepSeek, Qwen, Llama, and Mistral—as a managed turnkey service on sovereign NVIDIA GPU infrastructure in Turkey. Generate your API key and deploy in minutes. Choose the monthly tier matching your workload and benefit from transparent pay-as-you-go metering for overage.

Why FixCloud?

✓ OpenAI-compatible standard API
✓ 100% KVKK & GDPR Compliant, Turkey Datacenter (Equinix IL4)
✓ High performance powered by NVIDIA GH200
✓ Dedicated endpoint options

Have your own model?

We deploy and host your fine-tuned model on dedicated GPUs with custom support and private endpoints.

Request custom model →

Standard Models

10 Ready Models

Model	Provider	Type	Context	Girdi $/1M	Output $/1M	Action
GPT-OSS 120B	OpenAI	chat	128K	0,99	2,99	API kullan →
Qwen3.6 35B A3B	Alibaba	chat	128K	0,49	1,99	API kullan →
Gemma 4 26B A4B	Google	chat	64K	0,39	1,49	API kullan →
DeepSeek V4.1 Flash Max	DeepSeek	agent, coding	256K	1,49	3,49	API kullan →
GLM-4.5-Air	Z.ai	chat	64K	0,99	3,49	API kullan →
Mistral Medium 3.5	Mistral AI	chat + vision	64K	1,49	3,99	API kullan →
Llama 3.3 70B Instruct	Meta	chat	32K	0,99	1,99	API kullan →
Whisper Large v3 Turbo	OpenAI	speech to text	—	1,49	2,49	API kullan →
Qwen3 Embedding	Alibaba	embedding	32K	0,39	0,99	API kullan →
BGE Reranker v2	BAAI	reranking	8K	0,19	0,49	API kullan →

Token Metering: Service continues uninterrupted when your monthly package quota is reached; any additional volume is metered per 1M tokens at each model's native input/output rate and settled at the end of the billing period.

STANDARD PLANS

LLMaaS Plans Tailored to Your Needs

Generate your API key and launch in minutes. Transparent pay-as-you-go rates for usage beyond included quota.

PRO

For independent developers and starting projects.

$99 / mo

✓ Monthly token quota: 50 million tokens
✓ API key included
✓ All standard models
✓ Deep Search included
✓ Research Jobs: 10 / mo
✓ Batch Jobs included
✓ AI Studio & Compare
✓ Rate limit: 30 req/min
✓ Concurrent requests: 4
✓ Activity telemetry & export
✓ Monthly spending limit: Configurable
✓ Support: Standard

Start Pro plan →

BUSINESS

For growing teams and enterprise applications.

$199 / mo

✓ Monthly token quota: 100 million tokens
✓ API key included
✓ All standard models
✓ Deep Search included
✓ Research Jobs: 50 / mo
✓ Batch Jobs: Priority
✓ AI Studio & Compare
✓ Rate limit: 100 req/min
✓ Concurrent requests: 16
✓ Activity telemetry & export
✓ Monthly spending limit: Configurable
✓ Support: Priority

Start Business plan →

ULTRA

For high-volume enterprise production inference.

$499 / mo

✓ Monthly token quota: 500 million tokens
✓ API key included
✓ All standard models
✓ Deep Search included
✓ Research Jobs: Unlimited
✓ Batch Jobs: Priority
✓ AI Studio & Compare
✓ Rate limit: 500 req/min
✓ Concurrent requests: 64
✓ Activity telemetry & export
✓ Monthly spending limit: Configurable
✓ Support: Premium

Start Ultra plan →

Serverless vLLM Architecture

Serverless vLLM Inference with Your Own Models

Zero idle costs when traffic drops (scale-to-zero), sub-second cold starts on first request, and high token throughput powered by vLLM.

PagedAttention Architecture

Intelligent KV-cache management that eliminates memory fragmentation, delivering 2 to 4x higher throughput and lower latency.

Scale-to-Zero & Per-Second Billing

GPU resources automatically spin down to zero when idle. Pay strictly for the active seconds your models serve inference.

Continuous Batching

Dynamically groups incoming requests on the fly to minimize wait times and hardware cost per request.

Automated Model Network Cache

Popular weights like Llama 3.3, Mistral, Qwen, and DeepSeek are pre-cached locally in our Istanbul datacenter for instant spin-up.

Serverless vLLM'i başlat →

Data Sovereignty & Security

Your Data Belongs Strictly to You • 100% Sovereign Infrastructure

Process and store sensitive data locally with 100% KVKK and GDPR compliance without leaving national borders

100% KVKK Compliant

No enterprise data ever leaves national borders. All model training, inference, and storage are hosted strictly in our Istanbul datacenters.

Enterprise SIEM Log Forwarding

Real-time streaming via Syslog (UDP/TCP) and CEF directly to your internal enterprise SOC / SIEM systems.

AES-256 & TLS 1.3 Encryption

AES-256 at rest and TLS 1.3 in transit end-to-end encryption standards ensuring maximum enterprise cybersecurity.

What Our Clients Say

Real Success Stories with FixCloud AI

How do companies run their AI workloads on FixCloud without their data leaving Turkey? In our customers' own words.

Financial Assistant · LLMaaS + GPUaaS

HangiKredi took its Financial Assistant live without data leaving Turkey

HangiKredi, which compares financial products for millions of users, runs its agentic AI infrastructure and its Financial Assistant product on FixCloud LLMaaS and GPUaaS. Under KVKK and the Banking Law the data had to stay in the country; the team chose FixCloud on the performance results of tests and POCs it ran with many providers.

Read the full story →

Data stays in Turkey Processing in line with KVKK and the Banking Law
Live within a few months First inside the company, then opened to users
Predictable cost Fixed GPUaaS cost, no DevOps burden on the team

“It gave us the comfort of on-prem without having to bear the DevOps cost.”

Orkun Sevinç · HangiKredi, Data Management and AI

FAQ

Frequently Asked Questions

All the answers to your questions about the FixCloud.AI sovereign cloud platform

How does FixCloud.AI guarantee my data never leaves Turkey? ▼

All of your data is hosted and processed exclusively in our Tier-III/IV certified datacenters in Istanbul. Under KVKK Article 9 and GDPR standards, no data is ever transferred abroad. Network traffic connects directly through local Turkish peering exchanges.

How do I migrate my existing OpenAI API application to FixCloud.AI? ▼

FixCloud.AI endpoints are 100% drop-in compatible with OpenAI REST API standards. Simply update the base_url in your existing Python, Node.js, or cURL client to https://llm.fixcloud.ai/v1 to migrate in seconds without code rewrites.

Can I rent GPU servers in 1, 2, and 4-card physical configurations on an hourly basis? ▼

Yes. You can rent NVIDIA GH200 NVL2 288GB and RTX PRO 6000 Blackwell GPU servers in 1, 2, and 4-card physical bare-metal configurations with granular hourly pay-as-you-go billing or guaranteed monthly ceiling price advantages.

How does enterprise SIEM and log forwarding (Syslog / CEF) work? ▼

On Business and Ultra tiers, real-time API access and transaction logs are automatically streamed via Syslog (UDP/TCP) or CEF directly into your internal corporate SIEM / SOC behind your enterprise firewall.

What are the payment options? Can we be invoiced in Turkish Lira (TRY)? ▼

Yes. Invoices are officially issued in Turkish Lira (TRY) or USD for both individual and corporate accounts. We accept major credit cards, bank wire / EFT, and corporate commercial invoice billing accounts.

How do I sign up for FixCloud.AI? ▼

Sign up on the sign-up page with your company e-mail address; free e-mail providers such as Gmail or Hotmail are not accepted. After you verify the 6-digit code sent to your mobile phone by SMS, your first password and a verification link are sent to your e-mail, and you are asked to change the password at first sign-in.

Where do I get my API key and which endpoint should I use? ▼

An active LLMaaS plan is required to use the LLM API. Once you have chosen a plan, create your API key in the customer console and send your requests to the https://llm.fixcloud.ai/v1 endpoint.

What happens when my monthly token quota runs out? ▼

Your access is not cut off when the quota is used up; usage above the plan quota is billed separately at the token prices at the end of the period. The overage is capped by your wallet balance and by the monthly LLM budget you can set in the customer console; once that cap is reached, requests are rejected until you top up or raise the limit.

Where can I follow the live status of the platform? ▼

The Platform Status page shows the live state of the API, database, queue and LLM inference gateway. It is measured live and refreshes every 30 seconds; scheduled maintenance notices appear there too.