AI inference · billed in Naira
AI infrastructure that
bills in Naira.
OpenAI-compatible inference for Nigerian developers and product teams. Top up with your Nigerian card via Paystack, ship to your users in Lagos or Abuja, and stop paying twice for FX volatility.
Self-serve API keys in 60 seconds. No card required to start. Top up from ₦1,000.
from openai import OpenAI
client = OpenAI(
api_key="gpb_...",
base_url="https://api.gpubox.ai/v1",
)
# Same SDK as OpenAI. Same response shape. Bills in Naira at top-up.
response = client.chat.completions.create(
model="qwen2.5-32b-instruct",
messages=[{"role": "user", "content": "Summarise this in 2 lines: ..."}],
)
print(response.choices[0].message.content)
# usage: response.usage.total_tokens → e.g. 412 tokens ≈ ₦0.89Why Naira at the till
We charge you in Naira because that's how Nigerian businesses pay.
# OpenAI charges in
currency = "USD" # your bank gets a haircut
# GPUBox NG charges in
currency = "NGN" # Paystack, no FX surprise
Models
Same models, Naira-quoted rates
LLM · chat completions
liveqwen2.5-32b-instruct
Qwen2.5-32B-Instruct (AWQ-int4) on RTX 5090. 8k context. Streaming, tool use, JSON mode. ₦2,150 / 1M tokens.
/v1/chat/completions
Speech-to-text
livewhisper-large-v3-turbo
Whisper large-v3-turbo via faster-whisper. 100+ supported languages. ₦10.75 / audio minute. (Quality varies by language — benchmark on your own audio before production.)
/v1/audio/transcriptions
Embeddings
livebge-m3
BGE-M3 multilingual 1024-dim dense vectors, 8k context. ₦107.50 / 1M tokens.
/v1/embeddings
Why GPUBox for Nigerian builders
Naira-native checkout. Sovereign infrastructure. Honest pricing.
Top up in Naira, no FX runaround
Pay with your Nigerian card via Paystack. We invoice in NGN, settle in NGN, no story. The credit lands as GBP under the hood so your usage costs are predictable globally — but at the till, it's Naira.
OpenAI-compatible — change one URL
Every line of your existing OpenAI client code keeps working. Python, Node, Go, curl, Langchain, OpenAI Python SDK — point base_url at api.gpubox.ai/v1 and you're shipping.
UK-incorporated, NDPR-aware
GPUBox is operated by Mobile Paradigm Consultancy Ltd (UK, VAT GB397067846), trading as Frontendlabs. UK contract, English law — useful when enterprise procurement asks 'who is the counterparty'. NDPR processor schedule with IDTA / SCC transfer mechanism available on request. (Inference runs in the UK — if you need data-in-Nigeria today, see the sovereignty page first.)
Real model names, no silent swaps
Qwen2.5-32B-Instruct on chat. Whisper-large-v3-turbo on audio. BGE-M3 on embeddings. We never substitute one model for another to fit our margins — what you ask for is what runs.
Pay-as-you-go
Honest, published rates.
One blended rate per million tokens. One rate per audio minute. No GPU-hour roulette. No mystery bill.
See full pricing →Chat / LLM
₦2,150
per 1M tokens
≈ £1.00
Audio / Whisper
₦10.75
per audio minute
≈ £0.005
Built in the UK. Billed in Naira.
Sign up in 60 seconds. Top up from ₦1,000 via Paystack. Change one URL and ship.