:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
import clsx from 'clsx'
import { LemonTag } from '@posthog/lemon-ui'
import { dayjs } from 'lib/dayjs'
import { lowercaseFirstLetter } from 'lib/utils/strings'
import { CostBreakdownTooltip } from '../components/CostBreakdownTooltip'
import { MetadataTag } from '../components/MetadataTag'
import { CostContext, aiTokenCount, formatLLMCost, hasCostBreakdown } from '../utils'
export function MetadataHeader({
inputTokens,
outputTokens,
costContext,
model,
latency,
className,
isError,
cacheReadTokens,
cacheWriteTokens,
timestamp,
timeToFirstToken,
isStreaming,
}: {
/** Token props read leniently, so a string or `{ total, … }` count still renders its tag. */
inputTokens?: unknown
outputTokens?: unknown
cacheReadTokens?: unknown
cacheWriteTokens?: unknown
costContext?: CostContext
model?: string
latency?: number
isError?: boolean
className?: string
timestamp?: string
timeToFirstToken?: number
isStreaming?: boolean
}): JSX.Element {
const inputTokenCount = aiTokenCount(inputTokens)
const outputTokenCount = aiTokenCount(outputTokens)
const cacheReadTokenCount = aiTokenCount(cacheReadTokens)
const cacheWriteTokenCount = aiTokenCount(cacheWriteTokens)
return (
<div className={clsx('flex flex-wrap gap-2', className)}>
{isError && <LemonTag type="danger">Error</LemonTag>}
{typeof latency === 'number' && (
<MetadataTag
label="Latency"
tooltipContent="End-to-end latency from request to final response."
>{`${Math.round(latency * 10e2) / 10e2}s latency`}</MetadataTag>
)}
{typeof timeToFirstToken === 'number' && isStreaming && (
<MetadataTag
label="Time to first token"
tooltipContent="Time between sending the request and receiving the first streamed token."
>
{timeToFirstToken < 1
? `${Math.round(timeToFirstToken * 1000)} ms`
: `${timeToFirstToken.toFixed(2)}s TTFT`}
</MetadataTag>
)}
{timestamp && <MetadataTag label="Timestamp">{dayjs(timestamp).format('MMM D, YYYY h:mm A')}</MetadataTag>}
{inputTokenCount !== null && outputTokenCount !== null && (
<MetadataTag
label="Token usage"
tooltipContent="Input and output tokens consumed by this generation call."
>
{`${inputTokenCount} prompt tokens → ${outputTokenCount} completion tokens (∑ ${
inputTokenCount + outputTokenCount
})`}
</MetadataTag>
)}
{cacheReadTokenCount !== null && cacheReadTokenCount > 0 && (
<MetadataTag
label="Cache read"
tooltipContent="Prompt tokens served from the provider's cache (cheaper, faster)."
>{`${cacheReadTokenCount} cache read tokens`}</MetadataTag>
)}
{cacheWriteTokenCount !== null && cacheWriteTokenCount > 0 && (
<MetadataTag
label="Cache write"
tooltipContent="Prompt tokens written to the provider's cache for reuse on subsequent calls."
>{`${cacheWriteTokenCount} cache write tokens`}</MetadataTag>
)}
{typeof model === 'string' && model && (
<MetadataTag label="Model" textToCopy={lowercaseFirstLetter(model)}>
{model}
</MetadataTag>
)}
{costContext && (
<MetadataTag
label="Total generation cost"
tooltipContent={
hasCostBreakdown(costContext) ? <CostBreakdownTooltip costContext={costContext} /> : undefined
}
>
{formatLLMCost(costContext.totalCost)}
</MetadataTag>
)}
</div>
)
}