7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.
TypeScript
import { useI18n } from '@/i18n'
import { formatTokens, platformColors } from '@/lib/routing'
// Month-to-date spend for the embedding / image / audio tabs, in the same card
// slot the chat tab uses for its budget bar.
//
// It is deliberately NOT a budget bar. Chat can draw a percentage because
// `models.monthly_token_budget` is a real number. `embedding_models` and
// `media_models` only carry a free-text `quota_label` — "10K neurons/day
// (shared)", "$0.10/mo credits", "MP3 output - multilingual" — and there is no
// honest conversion from Neurons or dollar credits to tokens, nor is the last
// one a quota at all. So the bar shows how spend is *distributed* across
// providers (which is real), the total is what was actually spent, and each
// provider's quota label is printed verbatim for the reader to judge.
export interface UsageRow {
/** Family name for embeddings, model display name for media. */
label: string
platform: string | null
quotaLabel?: string | null
/** Whichever unit this modality is metered in. */
amount: number
requestsToday: number
}
export function UsageSummaryCard({
rows,
total,
requestsToday,
unit,
}: {
rows: UsageRow[]
total: number
requestsToday: number
/** 'tokens' for embeddings, 'requests' for image/audio (billed per image or per character). */
unit: 'tokens' | 'requests'
}) {
const { t } = useI18n()
const spent = rows.filter(r => r.amount > 0)
const fmt = (n: number) => (unit === 'tokens' ? formatTokens(n) : String(n))
return (
<section className="rounded-3xl border bg-card p-5">
<div className="flex items-baseline justify-between gap-3 mb-3">
<h2 className="text-sm font-medium">{t('models.usageThisMonth')}</h2>
<span className="text-xs text-muted-foreground tabular-nums">
{unit === 'tokens'
? t('embeddings.usageMonth', { count: fmt(total) })
: t('models.usageRequestsMonth', { count: total })}
<span className="mx-1.5">·</span>
{t('embeddings.usageToday', { count: requestsToday })}
</span>
</div>
{/* The track always renders. Hiding it at zero made the whole card read as
missing on tabs with no traffic yet, which is exactly when a reader is
checking whether the feature is there at all. */}
{total > 0 ? (
<div className="flex h-2.5 rounded-full overflow-hidden bg-muted">
{spent.map((r, i) => (
<div
key={i}
title={`${r.label}${r.platform ? ` (${r.platform})` : ''}: ${fmt(r.amount)}`}
style={{
width: `${(r.amount / total) * 100}%`,
backgroundColor: platformColors[r.platform ?? ''] ?? '#94a3b8',
}}
/>
))}
</div>
) : (
<div className="h-2.5 rounded-full bg-muted" />
)}
<div className="mt-4 grid grid-cols-1 sm:grid-cols-2 lg:grid-cols-3 gap-x-5 gap-y-1.5 text-xs tabular-nums">
{rows.map((r, i) => (
<div key={i} className="flex items-center gap-2 min-w-0">
<span
className="size-2 rounded-sm flex-shrink-0"
style={{ backgroundColor: platformColors[r.platform ?? ''] ?? '#94a3b8' }}
/>
<span className="truncate">{r.label}</span>
{r.quotaLabel && (
<span className="truncate text-muted-foreground/70 font-normal">{r.quotaLabel}</span>
)}
<span className="flex-1" />
<span className="font-mono text-muted-foreground">{fmt(r.amount)}</span>
</div>
))}
</div>
{total === 0 && (
<p className="mt-3 text-xs text-muted-foreground">{t('models.usageEmpty')}</p>
)}
</section>
)
}