Aggregate production metrics for fikraapi.co.ke — every POST /v1/chat/completions (and embeddings/audio) call that settles a bill is counted; upstream faults are logged and exposed below. No user or key data is shown.
| Incident | Severity | Status | Window (UTC) | Downtime in window |
|---|---|---|---|---|
| Model catalog, pricing and inference outage | MAJOR | RESOLVED | 2026-10-18 13:20 → 2026-10-18 13:42 | 22.0 min |
Total tokens across all models, per UTC day (successes only; the ledger is add-as-you-succeed).
| Model | POST requests | Tokens | Uptime (fault-adjusted) | Status |
|---|---|---|---|---|
| fikra-pro-120b | 10,075 | 26M | 99.921% (8 faults) | degraded recently |
| fikra-qwen-max | 850 | 3M | 100% (no faults logged yet) | up |
| fikra-flash | 814 | 1M | 100% (no faults logged yet) | up |
| fikra-pro-20b | 769 | 1M | 100% (no faults logged yet) | up |
| nvidia-nemotron-3.5-super-vl | 50 | 1M | 100% (no faults logged yet) | up |
| kimi-k3 | 23 | 0M | 100% (no faults logged yet) | up |
| step-5-preview | 24 | 0M | 100% (no faults logged yet) | up |
| deepseek-v4.1-flash | 125 | 0M | 100% (no faults logged yet) | up |
| fikra-code-35b | 60 | 0M | 100% (no faults logged yet) | up |
| fikra-qwen-27b | 69 | 0M | 100% (no faults logged yet) | up |
| qwen3.8-27b | 64 | 0M | 100% (no faults logged yet) | up |
| glm-5.3-flash | 80 | 0M | 100% (no faults logged yet) | up |
| fikra-auto | 50 | 0M | 100% (no faults logged yet) | up |
| fikra-fast-8b | 10 | 0M | 100% (no faults logged yet) | up |
| fikra-auto-pro | 6 | 0M | 100% (no faults logged yet) | up |
| fikra-pro-21b | 1 | 0M | 100% (no faults logged yet) | up |
| deepseek-v4-flash-0731 | 3 | 0M | 100% (no faults logged yet) | up |
| fikra-embed | 7 | 0M | 100% (no faults logged yet) | up |
| fikra-free-deepseek | 2 | 0M | 100% (no faults logged yet) | up |
| glm-5.3 | 1 | 0M | 100% (no faults logged yet) | up |
| deepseek-v4.1-flash-1017 | 1 | 0M | 100% (no faults logged yet) | up |
| fikra-tts | 8 | — | 100% (no faults logged yet) | up |
| fikra-stt | 5 | — | 100% (no faults logged yet) | up |
The open-frontier families customers can call today, alongside Fikra's core aliases. Volume and health per model over the window.
| Model | Family | Requests (30d) | Tokens | Status |
|---|---|---|---|---|
| deepseek-v4.1-flash | deepseek | 125 | 0M | live |
| deepseek-v4.1-flash-1017 | deepseek | 1 | 0M | live |
| deepseek-v4-flash-0731 | deepseek | 3 | 0M | live |
| fikra-free-deepseek | deepseek | 2 | 0M | live |
| glm-5.3-flash | glm | 80 | 0M | live |
| kimi-k3 | kimi | 23 | 0M | live |
| step-5-preview | stepfun | 24 | 0M | live |
| glm-5.3 | glm | 1 | 0M | live |
| glm-5.2 | glm | 0 | — | live |
Error funnels grouped by model + HTTP status, and the 20 most recent provider-side faults. 429s shown separately as they reflect per-account limits, not platform downtime. Error capture went live with this page; older errors were not instrumented.
| Model | HTTP | Count (30d) | |
|---|---|---|---|
| fikra-pro-120b | 429 | 8 | rate limit (account tier) |
| platform | 500 | 4 | provider fault |
| incident | 500 | 1 | provider fault |
| agent-error | 503 | 1 | provider fault |
| When (UTC) | Model | HTTP | Detail | Upstream |
|---|---|---|---|---|
| 2026-10-18T17:04:03Z | platform | 500 | TemplateSyntaxError: expected token 'end of print statement', got ':' | — |
| 2026-10-18T16:54:41Z | platform | 500 | RemoteProtocolError: Server disconnected | — |
| 2026-10-18T16:21:05Z | platform | 500 | RemoteProtocolError: Server disconnected | — |
| 2026-10-18T16:07:57Z | fikra-pro-120b | 429 | non-stream upstream fault key-cycle: {"error":{"message":"Model busy, retry later","type":"invalid_request_error","param":null,"code":"engine_overloaded"}} | deepinfra |
| 2026-10-18T16:07:49Z | fikra-pro-120b | 429 | non-stream upstream fault key-cycle: {"error":{"message":"Model busy, retry later","type":"invalid_request_error","param":null,"code":"engine_overloaded"}} | deepinfra |
| 2026-10-18T16:07:40Z | fikra-pro-120b | 429 | non-stream upstream fault key-cycle: {"error":{"message":"Model busy, retry later","type":"invalid_request_error","param":null,"code":"engine_overloaded"}} | deepinfra |
| 2026-10-18T15:32:50Z | platform | 500 | RemoteProtocolError: Server disconnected | — |
| 2026-10-18T14:58:00Z | fikra-pro-120b | 429 | non-stream upstream fault key-cycle: {"error":{"message":"Model busy, retry later","type":"invalid_request_error","param":null,"code":"engine_overloaded"}} | deepinfra |
| 2026-10-18T14:55:49Z | fikra-pro-120b | 429 | non-stream upstream fault key-cycle: {"error":{"message":"Model busy, retry later","type":"invalid_request_error","param":null,"code":"engine_overloaded"}} | deepinfra |
| 2026-10-18T14:54:39Z | fikra-pro-120b | 429 | non-stream upstream fault key-cycle: {"error":{"message":"Model busy, retry later","type":"invalid_request_error","param":null,"code":"engine_overloaded"}} | deepinfra |
| 2026-10-18T14:53:01Z | fikra-pro-120b | 429 | non-stream upstream fault key-cycle: {"error":{"message":"Model busy, retry later","type":"invalid_request_error","param":null,"code":"engine_overloaded"}} | deepinfra |
| 2026-10-18T14:48:43Z | fikra-pro-120b | 429 | non-stream upstream fault key-cycle: {"error":{"message":"Model busy, retry later","type":"invalid_request_error","param":null,"code":"engine_overloaded"}} | deepinfra |
| 2026-10-18T13:40:25Z | incident | 500 | OUTAGE 13:20-13:42 UTC: models/feed/benchmark RLS enabled with zero policies -> all catalog reads returned empty (/, /models, /pricing, /v1/models), status page showed inference do | fikra-internal |
| 2026-10-18T13:40:25Z | agent-error | 503 | Agent-caused: benchmark/model-scout jobs created state that coincided with RLS lockdown; playground chats 503 (resolve_demo_model found 0 eligible models); top-up initialize unpubl | fikra-internal |