Every Mistral call, costed and tracked.
Mistral's models — Large, Small, Codestral, and the open weights — power your features; Obsivara tells you what each call costs and how reliably it runs. Every completion, embedding, and tool call traced with tokens, latency, and dollars attached, and compared against your other providers on real traffic. Instrument via the Obsivara SDK or OpenTelemetry, or POST one usage event per call — nothing routes through Obsivara and no latency is added.
Mistral, fully observable.
Per-call cost and token accounting
Input and output tokens and the exact dollar cost of every request, priced per Mistral model and attributed to the workflow and team that made it.
Latency and error monitoring
P50/P95/P99 latency per model and endpoint, rate-limit tracking, and error classification — so you know when Mistral is slow versus when your prompts are.
Model comparison on live traffic
See what moving a workload between Mistral tiers, or between Mistral and another provider, would save — measured on your actual requests.
Usage anomaly detection
Token spikes, runaway retries, and prompt-size creep flagged before they land on your invoice.
Three steps. No code changes.
Add your Mistral usage source
Instrument your app with the Obsivara SDK or an OpenTelemetry GenAI exporter, or POST one usage event per request to the ingest webhook. Your app keeps calling Mistral directly — Obsivara records each call out-of-band, so nothing routes through us and no latency is added.
Obsivara maps your usage
Within minutes you get a live inventory of every Mistral model in use, the workflows calling it, and what each costs.
Set budgets and alerts
Define spend thresholds, latency SLOs, and error-rate alerts per workflow. Obsivara watches from there.
Signals tracked out of the box.
- →Tokens per request (input / output)
- →Cost per call, workflow, and day
- →Latency P50 / P95 / P99 by model
- →Rate-limit and quota errors
- →Model and parameter drift between deploys
- →Prompt size creep over time
Common questions.
Does Obsivara need my Mistral API key?
No. Obsivara captures usage from the events your app sends via the SDK, OpenTelemetry, or a webhook, so it never needs your production keys and never calls Mistral on your behalf.
Will this add latency to my Mistral calls?
No. Obsivara never sits in your request path — your app calls Mistral directly and reports usage afterward. There's no proxy, so there's no added latency.
Does it cover Codestral and embedding models?
Yes. Chat, code, and embedding models are all tracked with model-accurate pricing and rolled up per workflow.