Claude API base token rates. Batch discounts, cache writes and plan-specific conditions are separate. Cached input price: $1.00 / 1M tokens (base tier).
Published billing units · original currency · observed 12 Oct 2026
Category
Usage
Rate
Unit
Text tokens · Base rate
Cache write (5m)1
$12.50
1M tokens
Text tokens · Base rate
Cache write (1h)1
$20.00
1M tokens
Rate conditions
Claude API standard rates; Base rate. Write rate replaces ordinary input billing for those tokens. Output and uncached input are separate. Batch, residency and service modifiers are not included.
Perplexity API key; POST https://api.perplexity.ai/v1/agent (compatibility alias: /v1/responses). Set service_tier=default. auto or an omitted tier also selects default processing. Anthropic requests require max_output_tokens.
Perplexity Agent API default processing; USD per million tokens. Tool calls are charged separately when used. Rates are host-specific, not the model developer’s direct-API prices. Unknown cache-write charges are not treated as zero. Cached input price: $1.00 / 1M tokens (base tier).
Published billing units · original currency · observed 12 Oct 2026
Category
Usage
Rate
Unit
Tool add-on
Web search1
$0.0025
invocation
Tool add-on
Fast search2
$0.001
invocation
Tool add-on
Image search3
$0.0025
successful invocation
Tool add-on
URL fetch4
$0.0005
invocation
Tool add-on
People search5
$0.005
invocation
Tool add-on
Finance search6
$0.005
invocation
Tool add-on
Code sandbox7
$0.03
container session
Tool add-on
Search inside sandbox8
$0.0025
SDK search request
Rate conditions
Standard web search. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
web_search with search_type=fast. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
Image search bills successful invocations. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
Fetch and extract URL content. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
People search tool. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
Financial search tool. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
Up to 20 minutes of active use per billing window; this is not a runtime cap. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
Additional SDK search calls made inside a sandbox. Optional Agent API tool usage is separate from token billing. This fee is not proof that this model supports every tool. Token service-tier multipliers are not applied to these tool fees.
Set top-level cache_control for automatic caching or place it on content blocks. Up to four breakpoints; longer-lived blocks must precede shorter-lived blocks.Source for How to enable it (opens in a new tab)
5m writes: 1.25× base input. 1h writes: 2×. Reads: 0.025× for Fable/Mythos 5.1, 0.05× for Opus/Sonnet 5.5, otherwise 0.1×. Host prices, long-context tiers and residency modifiers still apply.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Prefix order is tools → system → messages. Cache must exist before concurrent requests can reuse it. Cache isolation is per organization, and also per workspace on Claude API, Claude Platform on AWS and Foundry.Source for Limits & data boundaries (opens in a new tab)
Claude API base token rates. Batch discounts, cache writes and plan-specific conditions are separate.
Claude API standard rates; Base rate. Write rate replaces ordinary input billing for those tokens. Output and uncached input are separate. Batch, residency and service modifiers are not included.
Observed 12 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.
Use the model’s Converse cachePoint or InvokeModel cache_control fields. Legacy InvokeModel does not accept top-level automatic cache_control; use explicit blocks.Source for How to enable it (opens in a new tab)
Usage fields
Converse: cacheReadInputTokens, cacheWriteInputTokens and cacheDetails. Native Messages responses use cache_read_input_tokens and cache_creation_input_tokens.Source for Usage fields (opens in a new tab)
Set top-level cache_control for automatic caching or place it on content blocks. Up to four breakpoints; longer-lived blocks must precede shorter-lived blocks.Source for How to enable it (opens in a new tab)
5m writes: 1.25× base input. 1h writes: 2×. Reads: 0.025× for Fable/Mythos 5.1, 0.05× for Opus/Sonnet 5.5, otherwise 0.1×. Host prices, long-context tiers and residency modifiers still apply.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Prefix order is tools → system → messages. Cache must exist before concurrent requests can reuse it. Cache isolation is per organization, and also per workspace on Claude API, Claude Platform on AWS and Foundry.Source for Limits & data boundaries (opens in a new tab)
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Set top-level cache_control for automatic caching or place it on content blocks. Up to four breakpoints; longer-lived blocks must precede shorter-lived blocks.Source for How to enable it (opens in a new tab)
5m writes: 1.25× base input. 1h writes: 2×. Reads: 0.025× for Fable/Mythos 5.1, 0.05× for Opus/Sonnet 5.5, otherwise 0.1×. Host prices, long-context tiers and residency modifiers still apply.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Prefix order is tools → system → messages. Cache must exist before concurrent requests can reuse it. Cache isolation is per organization, and also per workspace on Claude API, Claude Platform on AWS and Foundry.Source for Limits & data boundaries (opens in a new tab)
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Set top-level cache_control for automatic caching or place it on content blocks. Up to four breakpoints; longer-lived blocks must precede shorter-lived blocks.Source for How to enable it (opens in a new tab)
5m writes: 1.25× base input. 1h writes: 2×. Reads: 0.025× for Fable/Mythos 5.1, 0.05× for Opus/Sonnet 5.5, otherwise 0.1×. Host prices, long-context tiers and residency modifiers still apply.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Prefix order is tools → system → messages. Cache must exist before concurrent requests can reuse it. Cache isolation is per organization, and also per workspace on Claude API, Claude Platform on AWS and Foundry.Source for Limits & data boundaries (opens in a new tab)
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Do not inherit the model developer’s direct-API retention policy onto the Perplexity endpoint.
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Published discounted cache-read rates apply to this default service tier. Write charges, minimum cacheable input, retention and cache controls have not been verified for this host.Source for Billing rules (opens in a new tab)
Perplexity Agent API default processing; USD per million tokens. Tool calls are charged separately when used. Rates are host-specific, not the model developer’s direct-API prices. Unknown cache-write charges are not treated as zero.
Observed 12 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.
Caching reviewed 12 Oct 2026
What the measurements say
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Proprietary Claude release with developer-operated and documented partner access. No public model weights are supplied in the release documentation.
