BroadRouter

Anthropic API pricing

Updated

Anthropic lists standard per-token prices for 18 models. The cheapest is Claude Haiku 5.5 at USD 0.10 input and USD 0.50 output per 1M tokens; the dearest is Claude Mythos 5.1 at USD 10 / USD 50. It publishes cached-input prices for all of them, plus batch and fast tiers (no flex or priority). Checked 2026-11-14.

Anthropic is a model maker: 18 of the 18 models here are its own. By maker: Anthropic 18. 41 prices, each read from Anthropic's own pricing page, checked 2026-11-14. BroadRouter is BroadNet's own site and does not sell Anthropic's services. Automated re-check 2026-11-16: 41 of 41 prices still matched the provider's page.

Models with a standard price
18
Median input / output
USD 5 / USD 25
Cached-input prices
18 of 18
Cheapest host for
15 of 15 shared models

Which models is Anthropic the cheapest provider for?

Of the 15 models that Anthropic and at least one other provider both price at the standard tier, Anthropic has the lowest price for 15, ranked on a blend of 3 input tokens to 1 output token (all of them tied with another provider at the same price; the maker's own API is listed first on a tie). Another 3 models have a standard price from Anthropic only.

Models where Anthropic is the cheapest provider
ModelAnthropic input / outputNext cheapestSaving on 3:1 blendProviders
Claude Opus 5.5USD 4 / USD 20DeepInfra USD 4 / USD 20tie6
Claude Sonnet 5.5USD 2 / USD 10DeepInfra USD 2 / USD 10tie6
Claude Fable 5USD 10 / USD 50DeepInfra USD 10 / USD 50tie5
Claude Haiku 4.5USD 1 / USD 5DeepInfra USD 1 / USD 5tie5
Claude Opus 4.7USD 5 / USD 25DeepInfra USD 5 / USD 25tie5
Claude Opus 4.8USD 5 / USD 25DeepInfra USD 5 / USD 25tie5
Claude Opus 5USD 5 / USD 25DeepInfra USD 5 / USD 25tie5
Claude Sonnet 4.6USD 3 / USD 15DeepInfra USD 3 / USD 15tie5
Claude Sonnet 5USD 2 / USD 10DigitalOcean USD 2 / USD 10tie5
Claude Fable 5.1USD 10 / USD 50DigitalOcean USD 10 / USD 50tie4
Claude Fable 5.2USD 8 / USD 40DigitalOcean USD 8 / USD 40tie4
Claude Haiku 5.5USD 0.10 / USD 0.50DigitalOcean USD 0.10 / USD 0.50tie4
Claude Opus 4.5USD 5 / USD 25DigitalOcean USD 5 / USD 25tie4
Claude Opus 4.6USD 5 / USD 25DigitalOcean USD 5 / USD 25tie4
Claude Sonnet 4.5USD 3 / USD 15DigitalOcean USD 3 / USD 15tie4

The 6 models Anthropic sells that the most other providers also sell, with Anthropic's rank among them on the 3:1 blend.

Anthropic on the most widely hosted models
ModelAnthropicRankCheapest
Claude Opus 5.5USD 4 / USD 201 of 6Anthropic (this page)
Claude Sonnet 5.5USD 2 / USD 101 of 6Anthropic (this page)
Claude Fable 5USD 10 / USD 501 of 5Anthropic (this page)
Claude Haiku 4.5USD 1 / USD 51 of 5Anthropic (this page)
Claude Opus 4.7USD 5 / USD 251 of 5Anthropic (this page)
Claude Opus 4.8USD 5 / USD 251 of 5Anthropic (this page)

Anthropic price list

USD per 1M tokens: 41 prices for 18 models, one row per model with any other service tiers stacked under the standard price. Sources checked 2026-11-14 unless a row says otherwise. "vs maker" compares a standard price with the model maker's own API on the 3:1 blend; a minus means Anthropic is cheaper. Anthropic is the maker of every model here, so there is no maker price to compare with.

Anthropic API prices by model
ModelTierInputCachedOutputContextvs makerSource
Claude Fable 5standard
batch
USD 10
USD 5
USD 1
—
USD 50
USD 25
1,000,000own modelsource
Claude Fable 5.1standard
batch
USD 10
USD 5
USD 0.25
—
USD 50
USD 25
1,000,000own modelsource
Claude Fable 5.2standard
batch
fast
USD 8
USD 4
USD 16
USD 0.20
—
—
USD 40
USD 20
USD 80
1,000,000own modelsource
Claude Haiku 4.5standard
batch
USD 1
USD 0.50
USD 0.10
—
USD 5
USD 2.50
not statedown modelsource
Claude Haiku 5.5standard
batch
USD 0.10
USD 0.05
USD 0.01
—
USD 0.50
USD 0.25
not statedown modelsource
Claude Mythos 5standard
batch
USD 10
USD 5
USD 1
—
USD 50
USD 25
1,000,000own modelsource
Claude Mythos 5.1standard
batch
USD 10
USD 5
USD 0.25
—
USD 50
USD 25
1,000,000own modelsource
Claude Mythos 5.2standard
batch
fast
USD 8
USD 4
USD 16
USD 0.20
—
—
USD 40
USD 20
USD 80
1,000,000own modelsource
Claude Opus 4.5standard
batch
USD 5
USD 2.50
USD 0.50
—
USD 25
USD 12.50
not statedown modelsource
Claude Opus 4.6standard
batch
USD 5
USD 2.50
USD 0.50
—
USD 25
USD 12.50
1,000,000own modelsource
Claude Opus 4.7standard
batch
USD 5
USD 2.50
USD 0.50
—
USD 25
USD 12.50
1,000,000own modelsource
Claude Opus 4.8standard
batch
fast
USD 5
USD 2.50
USD 10
USD 0.50
—
—
USD 25
USD 12.50
USD 50
1,000,000own modelsource
Claude Opus 5standard
batch
fast
USD 5
USD 2.50
USD 10
USD 0.50
—
—
USD 25
USD 12.50
USD 50
1,000,000own modelsource
Claude Opus 5.5standard
batch
fast
USD 4
USD 2
USD 8
USD 0.20
—
—
USD 20
USD 10
USD 40
1,000,000own modelsource
Claude Sonnet 4.5standard
batch
USD 3
USD 1.50
USD 0.30
—
USD 15
USD 7.50
not statedown modelsource
Claude Sonnet 4.6standard
batch
USD 3
USD 1.50
USD 0.30
—
USD 15
USD 7.50
1,000,000own modelsource
Claude Sonnet 5standard
batch
USD 2
USD 1
USD 0.20
—
USD 10
USD 5
1,000,000own modelsource
Claude Sonnet 5.5standard
batch
USD 2
USD 1
USD 0.10
—
USD 10
USD 5
1,000,000own modelsource
  • Cached "—" means Anthropic's page shows no cache price for that row; budget cached tokens at the input price.
  • Context "not stated": the source does not give it (33 of 41 rows have one).

Report a wrong price or number (email to info@broadnet.ai, with this page's link filled in).

What service tiers does Anthropic publish?

  • batch (18 rows). Across 18 models with both prices, the median batch price is 50% below standard. Anthropic row note: “Message Batches API.”
  • fast (5 rows). Across 5 models with both prices, the median fast price is 2× standard. Anthropic row note: “Fast mode. Prompt caching multipliers apply on top; not available with the Batch API.”

Cached input: 18 of 18 standard prices include a cache-read price, a median 10% of the input price. What cached input means.

What we recorded about Anthropic's pricing page: Claude API (first-party), global routing (the default). US-only inference (inference_geo) costs 1.1x on Claude 4.6 and later. Batch and fast-mode cached-input prices are multipliers, not published figures, so they are null here. Retired models (Claude Opus 4.1, Opus 4, Sonnet 4) are left out. Context is set to 1M only where the page says Claude 4.6 and later models (except Haiku 5.5) include the full 1M token context window.

BroadNet Inference and Anthropic

BroadRouter is BroadNet's own site, so BroadNet is never ranked on this page. BroadNet Inference serves none of the models Anthropic prices. BroadNet Inference pricing.

More on Anthropic

Quick answers

Is Anthropic OpenAI-compatible?

Partly. Anthropic offers an OpenAI SDK compatibility layer for chat completions but says it is "primarily intended to test and compare model capabilities", not a long-term production path. It ignores strict on tools and returns 400 for audio and file inputs (Anthropic, OpenAI SDK compatibility, checked 2026-11-15).

What is Anthropic's API base URL?

https://api.anthropic.com/v1/, as Anthropic documents it. This is the base URL for the OpenAI SDK compatibility layer (Anthropic, OpenAI SDK compatibility, checked 2026-11-15).

Does Anthropic keep API prompts and outputs?

For up to 30 days. Anthropic says it automatically deletes API inputs and outputs "within 30 days of receipt or generation", except for features with longer retention under your control (such as the Files API), zero data retention agreements, Usage Policy enforcement and legal requirements. Flagged content may be kept for up to 2 years (Anthropic Privacy Center, commercial API data, checked 2026-11-14).

Does Anthropic train on API data?

No. Anthropic's Commercial Terms say: "Anthropic may not train models on Customer Content from Services" (Anthropic Commercial Terms of Service, checked 2026-11-15).

Does Anthropic have a free tier?

No free tier or free credit stated on the rate-limits page. Default limits (Start (monthly spend cap 500 USD), Claude Haiku 5.5): 4,000 requests a minute and 2,000,000 tokens a minute. Start tier; TPM shown is input tokens per minute (ITPM, uncached only), with 400,000 output tokens per minute. New organizations may start in an Evaluation tier with lower limits (Claude API rate limits, checked 2026-11-14).

Does Anthropic have a batch API?

Yes. Anthropic's Message Batches API gives a 50% discount, takes up to 100,000 requests or 256 MB per batch, and expires batches after 24 hours (Anthropic Message Batches API docs, checked 2026-11-15).

Questions

How much does the Anthropic API cost per token?

Anthropic's standard prices run from USD 0.10 input and USD 0.50 output per 1M tokens (Claude Haiku 5.5) to USD 10 / USD 50 (Claude Mythos 5.1). The median across 18 standard prices is USD 5 input and USD 25 output per 1M, which is USD 0.000005 per input token. Input, cached input and output are billed separately.

What is the cheapest model on Anthropic?

Claude Haiku 5.5, at USD 0.10 input and USD 0.50 output per 1M tokens (USD 0.20 on a 3:1 input-to-output blend).

Is Anthropic cheaper than other providers?

It depends on the model. Of 15 models that other providers also price, Anthropic is the cheapest standard price for 15 (all of them tied with another provider), for example Claude Opus 5.5, Claude Sonnet 5.5 and Claude Fable 5; another provider is cheaper for none; and it is the dearest for none. Its median rank is 1 of a median 5 providers, on a 3:1 input-to-output blend.

Does Anthropic have batch, cached or priority pricing?

Cached input: yes, for 18 of 18 standard prices, at a median 10% of the input price. Other tiers: batch on 18 rows (median input 50% below standard); fast on 5 rows (median input 2× standard).

How much does a chatbot cost per month on Anthropic?

For 2,000 requests a day, 1,500 input tokens (60% cached), 300 output, over 30 days: USD 13.14 on Claude Haiku 5.5, USD 657.00 on Claude Opus 4.6 (the middle of Anthropic's list) and USD 1,314 on Claude Mythos 5. Try your own numbers.

Read as MarkdownAsk ChatGPTAsk ClaudeAsk Perplexity