ONE API FOR ANY MODEL

Catalog of the most popular open-weight LLMs

OpenAI-compatible LLMaaS API. 10+ open-weight foundation models, instant access, token-based billing, ultra-low latency in our Istanbul Tier-IV datacenter.

Quick Start

OpenAI-Compatible API

Switch to sovereign local inference with a single line change to your base_url.

curl https://llm.fixcloud.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $YOUR_API_KEY" \
  -d '{
    "model": "gpt-oss-120b",
    "messages": [{"role": "user", "content": "Merhaba! Nasılsın?"}]
  }'
Model Catalog

10+ Open-Weight Foundation Models

Each model offers a distinct balance of intelligence, speed, and cost efficiency. Select the right model for your workload.

OpenAI OpenAI ● Live

GPT-OSS 120B

CHAT COMPLETION EMBEDDING 32K

OpenAI open-weight 120B MoE model (MXFP4). Advanced general reasoning and complex logical inference capabilities.

VRAM Memory 88 GB
Architecture MoE MXFP4
Total Parameters 120B
Max Context 32.000 Token
Alibaba Alibaba ● Live

Qwen3.6 35B A3B

CHAT 32K MoE (3B Aktif)

Alibaba Qwen3.6 35B A3B (MoE, 3B active params) - official FP8 checkpoint, next-generation ultra-low-latency Qwen architecture.

VRAM Memory 46 GB
Architecture MoE FP8
Total Parameters 35B
Max Context 32.000 Token
Google Google ● Live

Gemma 4 26B A4B

CHAT 32K MoE (3.8B Aktif)

Google Gemma 4 26B A4B (MoE, 3.8B active params) - dynamic FP8 quantization, highly efficient next-gen Gemma architecture.

VRAM Memory 44 GB
Architecture MoE FP8
Total Parameters 26B
Max Context 32.000 Token
DeepSeek DeepSeek ● Live

DeepSeek V4 Flash

CHAT REASONING AGENT CODING 32K

DeepSeek V4 Flash, optimized for agentic workflows, code generation, multi-step reasoning, and low-latency enterprise inference.

VRAM Memory 260 GB
Architecture MoE
Total Parameters 236B
Max Context 32.000 Token
Z.ai Z.ai ● Live

GLM-4.5-Air

CHAT REASONING AGENT CODING 64K

Open-weight MoE foundation model engineered for agentic applications (106B total / 12B active params) with hybrid reasoning modes.

VRAM Memory 250 GB
Architecture MoE
Total Parameters 106B
Max Context 64.000 Token
Mistral AI Mistral AI ● Beta / Live

Mistral Medium 3.5

VISION REASONING CODING FP8 32K

Mistral Medium 3.5: multimodal dense 128B model optimized for advanced reasoning, code synthesis, vision analysis, and agentic workflows.

VRAM Memory 220 GB
Architecture Dense FP8
Total Parameters 128B
Max Context 32.000 Token
Meta Meta · llama-3.3 ● Live

Llama 3.3 70B Instruct

CHAT INSTRUCT 32K

Meta Llama 3.3 70B Instruct - FP8 dinamik kuantizasyon. 405B sınıfına yakın metin kalitesi sunan tek GPU yoğun (dense) model.

VRAM Memory 92 GB
Architecture Dense FP8
Total Parameters 70B
Max Context 32.000 Token
OpenAI OpenAI ● Live

Whisper Large v3 Turbo

SPEECH-TO-TEXT MULTILINGUAL TRANSCRIPTION

Speed-optimized speech-to-text (STT) model from OpenAI. Highly accurate near-real-time transcription across 99+ languages.

VRAM Memory 12 GB
Architecture Encoder-Decoder
Total Parameters 1.5B
Max Audio 25 MB / 60 sn
Alibaba Alibaba ● Beta / Live

Qwen3 Embedding 8B

EMBEDDING MULTILINGUAL 32K

Multilingual text embedding model (top ranked on MTEB). Fully OpenAI /v1/embeddings compatible; produces 4,096-dim dense vectors.

VRAM Memory 20 GB
Architecture Decoder-Only
Total Parameters 8B
Max Context 32.000 Token
BAAI BAAI ● Beta / Live

BGE Reranker v2 M3

RERANKER MULTILINGUAL 8K

Multilingual cross-encoder reranker for RAG pipelines. Scores and reranks query-document pairs via Cohere-compatible /v1/rerank endpoint.

VRAM Memory 4 GB
Architecture Cross-Encoder
Total Parameters 568M
Max Context 8.000 Token
Custom Solution

Deep Research Agent

Multi-step research assistant chaining GPT-OSS 120B and DeepSeek V4 Flash. Includes comprehensive reporting and citations.

Subscriptions

LLMaaS Subscription Plans

Flexible subscription plans for AI workloads from solo developers to enterprise platforms.

Free

For beginners & explorers

$0 / month
  • ✓ 500K tokens / week quota
  • ✓ Limited model access (3 models)
  • ✓ Standard priority queue
  • ✓ REST API access

Business

Small & medium teams

$299 / month
  • ✓ 200M tokens / month quota
  • ✓ Full access to all models
  • ✓ 300 RPM rate limit
  • ✓ 15 concurrent requests
  • ✓ Priority support (email + phone)
  • ✓ Custom model fine-tuning
  • ✓ Usage analytics dashboard

Ultra

Enterprise & high volume

$999 / month
  • ✓ 1B tokens / month quota
  • ✓ Full access to all models
  • ✓ 1,000 RPM rate limit
  • ✓ 60 concurrent requests
  • ✓ Dedicated support (priority SLA)
  • ✓ Custom model fine-tuning (GPU dahil)
  • ✓ Deep Research agent access
  • ✓ Dedicated datacenter tier
  • ✓ Corporate billing + custom MSA
SSS

Frequently Asked Questions About LLMaaS

How does the LLMaaS API work with open-weight models? ▼
FixCloud provides a drop-in OpenAI-compatible API. Simply change the base_url in your existing OpenAI SDK client to access all models instantly.
What happens when my token quota runs out? ▼
In case of quota overage, additional usage is billed on transparent pay-as-you-go per-million-token rates, or you can seamlessly upgrade to a higher tier.
Is my data shared with third parties? ▼
No. All models are hosted in our Istanbul Tier-IV datacenter. No data is ever shared with third parties or outside borders. 100% KVKK and GDPR compliant.
Can I pay in Turkish Lira (TRY)? ▼
Yes, official VAT-compliant e-invoices are issued in either Turkish Lira (TRY) or USD for both individual and corporate accounts.

Start using the API today

Sign up for our OpenAI-compatible LLMaaS API for free and start sending requests in minutes.

Open console → Talk to enterprise sales