LLM🌍 Arabic Native
ALLaM 7B
SDAIA-trained Arabic-first large language model
$0.18/min≈ 0.67 SAR
Live GPU Marketplace
Rent GPU Compute
On Demand
Browse live GPU providers in Saudi Arabia. No account needed to browse — pay per minute, cancel anytime.
Choose your next path
One structure across DCP pages so each team can route directly to the right flow.
Self-serve renter
Create a renter account and start API + container job runs.
Provider onboarding
Register GPU hardware, install daemon, and start heartbeat checks.
Enterprise intake
Open procurement, security, and rollout planning support.
Arabic model docs
Review ALLaM 7B, Falcon H1, JAIS 13B, and BGE-M3 support paths.
Live Reliability
🌍 Arabic Models
Optimized for the Middle East — Arabic-trained models ready to deploy
Deploy
LLM🌍 Arabic Native
JAIS 13B
G42 Arabic/English bilingual foundation model
$0.22/min≈ 0.82 SAR
LLM🌍 Arabic Excellent
Qwen 2.5 7B
Strong Arabic instruction-following and reasoning
$0.15/min≈ 0.56 SAR
LLM🌍 Arabic Native
Falcon H1 7B
TII Arabic-aware hybrid state-space model
$0.16/min≈ 0.60 SAR
1 USD = 3.75 SAR • Prices shown in SAR for reference
Per-Model Rate Card
Updated 2026-05-13Pay per token, billed in halala (1 SAR = 100 halala). Open-source models; you pay compute cost plus a small platform margin.
DRAFT pricing — final rates pending founder review.
| Model | Input | Output |
|---|---|---|
ALLaM 7B InstructArabic-first ALLaM-AI/ALLaM-7B-Instruct-previewArabic-first, SDAIA-trained, PDPL-compliant | 12 halala / 1K tok SAR/1M: 120.00 | 36 halala / 1K tok SAR/1M: 360.00 |
JAIS 13B ChatArabic-first inceptionai/jais-13b-chatBilingual AR/EN, Inception/G42, strong Khaleeji dialect | 22 halala / 1K tok SAR/1M: 220.00 | 66 halala / 1K tok SAR/1M: 660.00 |
Falcon H1 7B (Arabic)Arabic-first tiiuae/Falcon-H1-7B-InstructTII Abu Dhabi, hybrid attention, Arabic + EN fine-tune | 12 halala / 1K tok SAR/1M: 120.00 | 36 halala / 1K tok SAR/1M: 360.00 |
Qwen 2.5 7B InstructChat Qwen/Qwen2.5-7B-InstructStrong multilingual, 128K context, LoRA-friendly | 10 halala / 1K tok SAR/1M: 100.00 | 30 halala / 1K tok SAR/1M: 300.00 |
Llama 3 8B InstructChat meta-llama/Meta-Llama-3-8B-InstructMeta open-weights, broad ecosystem support | 10 halala / 1K tok SAR/1M: 100.00 | 30 halala / 1K tok SAR/1M: 300.00 |
Mistral 7B Instruct v0.2Chat mistralai/Mistral-7B-Instruct-v0.2Compact, fast, Apache-2.0 | 9 halala / 1K tok SAR/1M: 90.00 | 27 halala / 1K tok SAR/1M: 270.00 |
Nemotron Nano 4BChat nvidia/Nemotron-Mini-4B-InstructConsumer-tier, instant cold-start (<4s), great for edge | 6 halala / 1K tok SAR/1M: 60.00 | 18 halala / 1K tok SAR/1M: 180.00 |
Qwen 2 72B InstructReasoning Qwen/Qwen2-72B-InstructFlagship multilingual reasoning, 128K context | 110 halala / 1K tok SAR/1M: 1100.00 | 330 halala / 1K tok SAR/1M: 3300.00 |
Llama 3 70B InstructReasoning meta-llama/Meta-Llama-3-70B-InstructGPT-3.5-class quality, fully open | 105 halala / 1K tok SAR/1M: 1050.00 | 315 halala / 1K tok SAR/1M: 3150.00 |
Nemotron Super 70BReasoning nvidia/Llama-3.1-Nemotron-70B-Instruct-HFNVIDIA reasoning fine-tune, strong on code + math | 115 halala / 1K tok SAR/1M: 1150.00 | 345 halala / 1K tok SAR/1M: 3450.00 |
Qwen 2.5 Coder 7BCode Qwen/Qwen2.5-Coder-7B-InstructCode-tuned, 92+ languages, fill-in-middle | 10 halala / 1K tok SAR/1M: 100.00 | 30 halala / 1K tok SAR/1M: 300.00 |
DeepSeek Coder V2 LiteCode deepseek-ai/DeepSeek-Coder-V2-Lite-InstructMoE 16B, 2.4B active params — fast code completion | 11 halala / 1K tok SAR/1M: 110.00 | 33 halala / 1K tok SAR/1M: 330.00 |
Stable Diffusion XLMultimodal stabilityai/stable-diffusion-xl-base-1.0Billed per image: 8 halala / 1024×1024 image (≈ SAR 0.08) | — | per image |
BGE-M3 EmbeddingsEmbeddings BAAI/bge-m3Multilingual, dense + sparse + colbert; ideal for Arabic RAG | 3 halala / 1K tok SAR/1M: 30.00 | 0 halala / 1K tok SAR/1M: 0.00 |
BGE Reranker v2-m3Embeddings BAAI/bge-reranker-v2-m3Cross-encoder reranker; pairs with BGE-M3 for top-k | 4 halala / 1K tok SAR/1M: 40.00 | 0 halala / 1K tok SAR/1M: 0.00 |
Anchored against OpenRouter / Together AI / DeepInfra published rates, May 2026. Conversion at SAR/USD = 3.75.
1 SAR = 100 halala. Rates are halala per 1,000 tokens. "vs market" compares output rate to OpenRouter/Together AI/DeepInfra published prices for the same weights.
Loading...