Skip to content

fix(aio): price Anthropic cache writes by TTL #71323

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
richardsolomou merged 2 commits into from
Jul 28, 2026
Merged

fix(aio): price Anthropic cache writes by TTL #71323

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
db5b57d
fix(aio): price Anthropic cache writes by TTL
gouveags Jul 28, 2026
12daf0d
Merge branch 'master' into fix/anthropic-cache-ttl-pricing
richardsolomou Jul 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
39 changes: 38 additions & 1 deletion docs/onboarding/ai-observability/_snippets/generation-event.tsx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -424,7 +424,20 @@ export const GenerationEvent = (): JSX.Element => {
</td>
<td>
<p>
<em>(Optional)</em> Price per cached token write
<em>(Optional)</em> Price per cached token write. For custom Anthropic pricing, this
applies to both cache TTLs unless <code>$ai_cache_write_1h_token_price</code> is
set.
</p>
</td>
</tr>
<tr>
<td style={propertyColumnStyle}>
<code>$ai_cache_write_1h_token_price</code>
</td>
<td>
<p>
<em>(Optional)</em> Price per token written to Anthropic's 1-hour cache. Takes
precedence over <code>$ai_cache_write_token_price</code> for 1-hour writes.
</p>
</td>
</tr>
Expand Down Expand Up @@ -501,6 +514,30 @@ export const GenerationEvent = (): JSX.Element => {
<td>
<p>
<em>(Optional)</em> Number of tokens written to cache (Anthropic-specific)
<br />
When both TTL-specific counts are present, PostHog uses them instead of this
aggregate. The aggregate should equal their sum; if either count is missing, PostHog
uses the aggregate.
</p>
</td>
</tr>
<tr>
<td style={propertyColumnStyle}>
<code>$ai_cache_creation_5m_input_tokens</code>
</td>
<td>
<p>
<em>(Optional)</em> Number of tokens written to Anthropic's 5-minute cache
</p>
</td>
</tr>
<tr>
<td style={propertyColumnStyle}>
<code>$ai_cache_creation_1h_input_tokens</code>
</td>
<td>
<p>
<em>(Optional)</em> Number of tokens written to Anthropic's 1-hour cache
</p>
</td>
</tr>
Expand Down
48 changes: 48 additions & 0 deletions frontend/src/taxonomy/core-filter-definitions-by-group.json
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

2 changes: 2 additions & 0 deletions nodejs/src/ingestion/common/steps/event-preprocessing/validate-ai-event-tokens.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,6 +49,8 @@ const TOKEN_PROPERTIES = [
'$ai_reasoning_tokens',
'$ai_cache_read_input_tokens',
'$ai_cache_creation_input_tokens',
'$ai_cache_creation_5m_input_tokens',
'$ai_cache_creation_1h_input_tokens',
] as const

const AI_EVENT_TYPES = ['$ai_generation', '$ai_embedding', '$ai_span', '$ai_trace', '$ai_metric', '$ai_feedback']
Expand Down
2 changes: 2 additions & 0 deletions nodejs/src/ingestion/common/steps/event-preprocessing/validate-ai-event-tokens.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,8 @@ const TOKEN_PROPERTIES = [
'$ai_reasoning_tokens',
'$ai_cache_read_input_tokens',
'$ai_cache_creation_input_tokens',
'$ai_cache_creation_5m_input_tokens',
'$ai_cache_creation_1h_input_tokens',
] as const

/**
Expand Down
1 change: 1 addition & 0 deletions nodejs/src/ingestion/pipelines/ai/costs/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -134,6 +134,7 @@ export const processCost = (event: EventWithProperties): EventWithProperties =>
completion_token: event.properties['$ai_output_token_price'],
cache_read_token: event.properties['$ai_cache_read_token_price'],
cache_write_token: event.properties['$ai_cache_write_token_price'],
cache_write_1h_token: event.properties['$ai_cache_write_1h_token_price'],
request: event.properties['$ai_request_price'],
web_search: event.properties['$ai_web_search_price'],
},
Expand Down
87 changes: 87 additions & 0 deletions nodejs/src/ingestion/pipelines/ai/costs/input-costs.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -213,6 +213,93 @@ describe('calculateInputCost()', () => {
expectCostToBeCloseTo(result, 0.004275)
})

it.each([
{
name: '5-minute',
cacheCreationTokens: 100,
ttlProperties: {
$ai_cache_creation_5m_input_tokens: 100,
$ai_cache_creation_1h_input_tokens: 0,
},
expectedCost: '0.003375',
},
{
name: '1-hour',
cacheCreationTokens: 100,
ttlProperties: {
$ai_cache_creation_5m_input_tokens: 0,
$ai_cache_creation_1h_input_tokens: 100,
},
expectedCost: '0.0036',
},
{
name: 'mixed-TTL',
cacheCreationTokens: 300,
ttlProperties: {
$ai_cache_creation_5m_input_tokens: 100,
$ai_cache_creation_1h_input_tokens: 200,
},
expectedCost: '0.004575',
},
{
name: 'mixed-TTL numeric strings',
cacheCreationTokens: 300,
ttlProperties: {
$ai_cache_creation_5m_input_tokens: '100',
$ai_cache_creation_1h_input_tokens: '200',
},
expectedCost: '0.004575',
},
{
name: 'legacy aggregate-only',
cacheCreationTokens: 300,
ttlProperties: {},
expectedCost: '0.004125',
},
])('prices $name cache creation tokens', ({ cacheCreationTokens, ttlProperties, expectedCost }) => {
const event = createAnthropicTestEvent(1000, undefined, cacheCreationTokens, ttlProperties)

const result = calculateInputCost(event, ANTHROPIC_MODEL)

expect(result).toBe(expectedCost)
})

it.each([
{
name: '5-minute count only',
ttlProperties: { $ai_cache_creation_5m_input_tokens: 100 },
},
{
name: '1-hour count only',
ttlProperties: { $ai_cache_creation_1h_input_tokens: 200 },
},
])('uses the aggregate when the TTL breakdown has $name', ({ ttlProperties }) => {
const event = createAnthropicTestEvent(1000, undefined, 300, ttlProperties)

const result = calculateInputCost(event, ANTHROPIC_MODEL)

expect(result).toBe('0.004125')
})

it('uses the generic custom cache-write rate for both TTLs when no 1-hour rate is set', () => {
const customModel = createTestModel({
provider: 'custom',
cost: {
prompt_token: 0.000003,
completion_token: 0.000015,
cache_write_token: 0.000004,
},
})
const event = createAnthropicTestEvent(1000, undefined, 300, {
$ai_cache_creation_5m_input_tokens: 100,
$ai_cache_creation_1h_input_tokens: 200,
})

const result = calculateInputCost(event, customModel)

expect(result).toBe('0.0042')
})

it('uses 1.25x multiplier fallback for cache write when not defined', () => {
const modelWithoutCacheWrite = createTestModel({
model: 'claude-2',
Expand Down
36 changes: 30 additions & 6 deletions nodejs/src/ingestion/pipelines/ai/costs/input-costs.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,6 +39,14 @@ const usesInclusiveAnthropicInputTokens = (event: PluginEvent): boolean => {
return provider === 'gateway' && framework === 'vercel'
}

const hasNumericProperty = (event: PluginEvent, key: string): boolean => {
const value = event.properties?.[key]
return (
(typeof value === 'number' && Number.isFinite(value)) ||
(typeof value === 'string' && value.length > 0 && Number.isFinite(Number(value)))
)
}

export const resolveCacheReportingExclusive = (event: PluginEvent): boolean => {
if (!event.properties) {
return false
Expand Down Expand Up @@ -177,12 +185,28 @@ export const calculateInputCost = (event: PluginEvent, cost: ResolvedModelCost):
const cachedTextTokens = cacheReadTokens - cachedAudioInputTokens

if (matchProvider(event, 'anthropic')) {
const cacheWriteTokens = numericProperty(event, '$ai_cache_creation_input_tokens')

const writeCost =
cost.cost.cache_write_token !== undefined
? bigDecimal.multiply(cost.cost.cache_write_token, cacheWriteTokens)
: bigDecimal.multiply(bigDecimal.multiply(cost.cost.prompt_token, 1.25), cacheWriteTokens)
const aggregateCacheWriteTokens = numericProperty(event, '$ai_cache_creation_input_tokens')
const cacheWrite5mTokens = numericProperty(event, '$ai_cache_creation_5m_input_tokens')
const cacheWrite1hTokens = numericProperty(event, '$ai_cache_creation_1h_input_tokens')
const hasCacheWriteBreakdown =
hasNumericProperty(event, '$ai_cache_creation_5m_input_tokens') &&
hasNumericProperty(event, '$ai_cache_creation_1h_input_tokens')
const cacheWriteTokens = hasCacheWriteBreakdown
? cacheWrite5mTokens + cacheWrite1hTokens
: aggregateCacheWriteTokens

const cacheWrite5mRate = cost.cost.cache_write_token ?? bigDecimal.multiply(cost.cost.prompt_token, 1.25)
const cacheWrite1hRate =
cost.cost.cache_write_1h_token ??
(cost.provider === 'custom' && cost.cost.cache_write_token !== undefined
? cost.cost.cache_write_token
: bigDecimal.multiply(cost.cost.prompt_token, 2))
const writeCost = hasCacheWriteBreakdown
? bigDecimal.add(
bigDecimal.multiply(cacheWrite5mRate, cacheWrite5mTokens),
bigDecimal.multiply(cacheWrite1hRate, cacheWrite1hTokens)
)
: bigDecimal.multiply(cacheWrite5mRate, cacheWriteTokens)

const cacheReadCost =
cost.cost.cache_read_token !== undefined
Expand Down
Loading
Loading