Fikra vs ModelGates: access and checkout in Africa

Live numbers as of 2026-11-14

These figures come from the public stats snapshot generated on 2026-11-14 from backend ledgers. They show volume and reliability only; they do not prove any single route wins your workload. Run your own 50 prompts before you move budget.

What the routes cost right now

Prices are USD per 1M tokens, taken from the live registry snapshot dated 2026-11-14. The KES columns use the checkout rate of KES 133 = $1; your bank or card issuer may add its own conversion margin on top.

The catalogue is authoritative at /v1/models (the OpenAI-compatible listing) and at /models (the public card with context windows and rate limits). Check both before a production commitment; prices move with provider terms.

See the live list from your terminal: curl -s https://api.fikraapi.co.ke/v1/models -H "Authorization: Bearer YOUR_FIKRA_KEY"

Fikra vs ModelGates: the checkable facts, as of 2026-11-14

Every ModelGates cell above comes from its official site or docs checked on 2026-11-14, not from hearsay. Where the official source does not state something (for example an uptime number or a Kenyan payment rail), this page says so instead of filling the gap.

Rate limits for Fikra come from the live plan data dated 2026-11-14: all tiers list 3,000 rpm and 250,000 rpd.

What the Fikra wallet costs you in practice

Funding is prepaid in USD. Kenya checkout shows M-Pesa, Airtel Money and card; the minimum top-up is $5, charged as KES 136.99 and credited as KES 133 at the published KES 133 = $1 checkout rate. Every other African market is card-only at this checkout.

Calls debit the wallet per token at the rates on /models. A $5 minimum buys roughly 100M input tokens on deepseek-v4.1-flash-1017 at $0.05 per 1M, minus cache effects. Lite, Pro and Max are $10, $25 and $60 per 30 days; the balance is the hard cap, so cost overrun is structurally impossible.

1. KES checkout math is exact and published; elsewhere the issuer sets the conversion rate and margin.
2. No monthly invoice, no foreign-merchant decline loop once the wallet is funded.
3. Public telemetry at /stats lets a finance lead check volume and uptime without opening provider dashboards.

When ModelGates is the sensible choice

Pick ModelGates when the specific capability matters more than the checkout: OpenAI-compatible API; published per-model pricing; 10K requests per minute per API key. Verified at https://modelgates.ai/ on 2026-11-14.

Also pick ModelGates if you already hold a funded account there, your compliance review has cleared it, or the exact model revision you need is only served there. Check the model ID, price, context and retention terms on its official site before you commit production traffic.

Do not switch on the strength of a leaderboard screenshot, an influencer post or a launch discount. Score the route, not the narrative.

When the Fikra wallet is the sensible choice

1. You are in Kenya and want M-Pesa or Airtel Money at checkout, or you are elsewhere in Africa and want one USD card top-up instead of per-provider accounts.
2. You want one OpenAI-compatible base_url (https://api.fikraapi.co.ke/v1) for DeepSeek, GLM, Qwen, Kimi and Gemini-family routes, billed to one wallet.
3. You want pinned route names at /v1/models you can reproduce in staging and production.
4. You want /stats and /status as public evidence for a finance or compliance review.
5. You want wallet caps per key so no route can overspend.

Run the comparison yourself in one afternoon

Create a Fikra key, fund $5, and run the same 50 prompts (include 10 failure-shaped ones) through deepseek-v4.1-flash-1017 and through your ModelGates route. Log request ID, model, tokens, cache hits, cost and wall time.

Score accepted tasks per dollar and per minute from your own network. The route that survives your prompt set wins; this page cannot know your workload.

1. For text: start with deepseek-v4.1-flash-1017 ($0.05 input, $0.15 output per 1M; 1,048,576-token context, 393,216 max output).
2. For structured output: qwen3.8-27b ($0.045 input, $0.35 output; 1,001,536-token context; temperature 0).
3. For code-heavy loops: glm-5.3-flash ($0.1875 input, $0.625 output; 327,680-token context).
4. Escalate the hard 10% to deepseek-v4-pro-pinned only if flash routes fail your bar.

curl https://api.fikraapi.co.ke/v1/chat/completions \
  -H "Authorization: Bearer YOUR_FIKRA_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek-v4.1-flash-1017","messages":[{"role":"user","content":"Summarise this ticket in two lines."}]}'

Questions buyers ask before switching

Migration is a base_url and a model-ID change for any OpenAI-compatible client. Nothing else in your stack moves.

If you need a contractual SLA, Fikra self-serve is not it; ask support about enterprise terms first, and read the same question on the competitor's side before you assume it is covered there either.