Instanslar katalogi
Mavjud GPU tezlatkichlar
VERUM AI bulutli GPU infratuzilmasi: AI, LLM, mashinaviy o‘qitish va rendering uchun mos konfiguratsiyalar.
141 GB HBM3e flagmani
NVIDIA Hopper Ultra
NVIDIA H200 SXM5
141 GB
HBM3e (4.8 TB/s)
Protsessor va RAM32 vCPU • 480 GB System RAM
CUDA Cores16,896
Disk xotirasiYuqori tezlikdagi 4 TB NVMe
Quyidagilar uchun tavsiya etiladi:
- DeepSeek-R1 671B — to‘liq kontekst
- Llama 3.3 70B va 405B FP8
- Yuqori darajadagi korporativ LLM modellar
$4.67 / soat
VERUM AI tarifi
Maksimal tezlik
NVIDIA Hopper
NVIDIA H100 SXM5
80 GB
HBM3 (3.35 TB/s)
Protsessor va RAM16 vCPU • 250 GB System RAM
CUDA Cores16,896
Disk xotirasiYuqori tezlikdagi NVMe
Quyidagilar uchun tavsiya etiladi:
- LLM o‘qitish (Llama 3, DeepSeek-R1)
- FP8 / FP16 o‘qitish
- Korporativ taqsimlangan ML
$3.50 / soat
VERUM AI tarifi
Sinovdan o‘tgan klassika
NVIDIA Ampere
NVIDIA A100 SXM4
80 GB
HBM2e (2.0 TB/s)
Protsessor va RAM12 vCPU • 125 GB System RAM
CUDA Cores6,912
Disk xotirasiNVMe SSD
Quyidagilar uchun tavsiya etiladi:
- Klassik Deep Learning
- O‘rta hajmdagi modellarni qo‘shimcha o‘qitish
- Kompyuter ko‘rishi va audio
$1.81 / soat
VERUM AI tarifi
48 GB uchun optimal muvozanat
NVIDIA Ada Lovelace
NVIDIA L40S
48 GB
GDDR6 with ECC
Protsessor va RAM16 vCPU • 128 GB System RAM
CUDA Cores18,176
Disk xotirasiNVMe SSD
Quyidagilar uchun tavsiya etiladi:
- Katta kontekstli vLLM inferensi
- Multimodal modellar (tasvir va matn)
- Generativ sun’iy intellekt
$1.03 / soat
VERUM AI tarifi
Eng ko‘p tanlanadigan
NVIDIA Ada Lovelace
NVIDIA GeForce RTX 4090
24 GB
GDDR6X
Protsessor va RAM8 vCPU • 64 GB System RAM
CUDA Cores16,384
Disk xotirasiNVMe SSD
Quyidagilar uchun tavsiya etiladi:
- ComfyUI / Stable Diffusion / Flux
- Ollama / vLLM’ni lokal ishga tushirish
- 3D rendering (Blender, Unreal)
$0.44 / soat
VERUM AI tarifi