Fikra API is a Nairobi-based reseller and access layer for frontier Chinese models, DeepSeek, Qwen, GLM, Kimi, priced in USD per million tokens, contracted in Kenyan shillings, and billable to a single API key. The pilot turns thirty days of serious evaluation into a shipped integration.
# One key. Every Chinese model. from openai import OpenAI client = OpenAI( base_url="https://api.fikraapi.co.ke/v1", api_key="fk_live_…", ) # DeepSeek, Qwen, GLM, Kimi, Gemini… r = client.chat.completions.create( model="deepseek-v4.1-flash", messages=[{"role":"user","content":"给我们写个周报"}] )
Already doing 1B+ tokens per month? Enterprise volume is at a separate tier with dedicated capacity, reserved TPM, and a custom rate card. Talk to James →
Pilot accounts start at production limits. Upgrade to dedicated capacity once you outgrow the shared pool.
USD, one time. No recurring, no seats, no free credits.
Service fee for setup, production-weight accounts and 30-day same-day support. Pay via Paystack (USD card) or KES mobile money after we confirm the SOW.
Production-ready pilot window with dedicated support from day 1.
Pilot seats do not come with free credits. You fund the wallet yourself; minimum top-up $50. Every token billed at list price.
$1 of credit buys ~133M cached tokens on DeepSeek V4.1 Flash ($0.0075 / 1M).
Submit the form with your use case, volume estimate and contact. We review within one business day.
We scope: models, concurrency, regions, data policy, budget. You get a 1-paragraph SOW on WhatsApp or WeChat within one business day.
KES inline (M-Pesa/card) or Paystack (USD/cards). Both one-time, no recurring, no seats fee. Contract signed by you and James before we activate.
Key issued same business day with production limits. We pair to your first production calls until your first release passes.
Every detail here shapes the SOW and the pilot limits we provision. Be specific.
DeepSeek V4.1 Flash (cached $0.0075 / 1M input), Qwen 3.8 Max, GLM 5.3, Kimi K3, Gemini 4 Hydrogen. Pilot keys use the same catalogue and rates as production.
Yes. For data-in-country requirements we deploy a private inference residency on AWS, Alibaba Cloud, Tencent Cloud, or Huawei Cloud inside two weeks. Pricing on request.
99.5% uptime target, Sev-1 response inside 1 hour, 4-hour resolution, service credits. Casual support 08:00-18:00 EAT Mon-Fri.
Work email MX/A lookup; rejection of invalid/blocked domains. Payment only after you confirm we can serve you, and account is created only after you sign up.
Yes. Top up from $5, self-service or invoice. KES 133 = $1 with a 3% fee. No monthly seat fee, no minimums after pilot.