minirouter.sh

Explore

RankingsCompare model benchmarksAnalyticsTokens served, liveChangelogWhat shippedReleasesNew model dossiersStatusProvider health

Build

ToolsRead public pagesGuidesWorked cost mathsIntegrationsClient setupProvidersModels by publisherBecome a supplierOffer inference capacitySetupCheck a connection
MINIStake MINIEarn credits every second
Open navigation
Get a key

Anthropic

Claude Haiku 5.5

Haiku with adaptive reasoning. Built for summaries, extraction and focused subagent tasks.

Get a MiniRouter keyCompare this modelSee the evidence
Successor to Haiku 4.5Text requeststext + image inputtext output

Replaces Claude Haiku 4.5: input $1.05 → $0.105 per 1M (-90%), output $5.25 → $0.525 per 1M (-90%), context 200K → 1M, max output 64K → 128K.

Run this model

anthropic/claude-haiku-5.5

Use an endpoint supported by this model.

Get a key for this setup →
Thinking guide →
const response = await fetch('https://api.minirouter.sh/v1/chat/completions', {  method: 'POST',  headers: {    Authorization: `Bearer ${process.env.MINIROUTER_KEY}`,    'Content-Type': 'application/json',  },  body: JSON.stringify({    model: 'anthropic/claude-haiku-5.5',    messages: [{ role: 'user', content: 'Why is the sky blue?' }],    max_tokens: 1024,  }),}) const data = await response.json()console.log(data.choices[0].message.content)
Read docs →

Use text input and an explicit max_tokens limit. Remove temperature, top_p and top_k when migrating. Thinking uses output tokens; leave room for the answer. Native images, PDFs and computer-use tools are not covered by this text route.

Higher rates above 100K input tokensBase prices shown include MiniRouter's markup. Above 100,000 input tokens, the provider charges 5× the base input and output rates for the full request. Reasoning tokens count as output. The temporary credit hold covers conservative provider rates; final charges follow reported usage. Source ↗

Model overview

Input / 1M
$0.105
All-in MiniRouter rate
Output / 1M
$0.525
All-in MiniRouter rate
Context
1M
1,000,000 tokens
Max output
128K
tokens
Reasoning
Adaptive
Match effort to the task
Price threshold
100K input
Higher full-request rates above it
Announced
7 Oct 2026
publisher release
Focused work

Give Haiku a bounded task.

Anthropic positions Haiku 5.5 for everyday work at scale. Evaluate larger models for open-ended agentic coding.
Anthropic announcement
  1. [ 01 ]

    Extract and classify

    Define the fields or labels you need, then check outputs against representative examples.

  2. [ 02 ]

    Summarize supplied text

    Ask for references to the source material. Verify that important qualifiers survive the summary.

  3. [ 03 ]

    Delegate a small step

    Use a clear stopping condition and validate the result before passing it to another agent.

Publisher list prices

Watch the 100K boundary.

Above 100,000 input tokens, higher rates apply to the full request. These publisher prices exclude MiniRouter markup; current route prices appear above.

Up to 100K input · per 1M tokens

Input
$0.10
Output
$0.50
Cache read
$0.01

Above 100K input · per 1M tokens

Input
$0.50
Output
$2.50
Cache read
$0.05
Anthropic model overview ↗, October 7, 2026. Cache eligibility depends on the request and provider.
View chart data
Haiku 5.5 publisher prices per million tokens
RateUp to 100K inputAbove 100K input
Input$0.10$0.50
Output$0.50$2.50
Cache read$0.01$0.05
From Haiku 4.5

Recheck options and token budgets.

Haiku 5.5 defaults to adaptive thinking at medium effort. Anthropic reports roughly 30% more input tokens for the same text than Haiku 4.5.
Anthropic migration guide
  1. [ 01 ]

    Remove legacy settings

    Native requests reject manual thinking budgets, assistant prefills and non-default sampling parameters.

  2. [ 02 ]

    Budget for reasoning

    Small output limits can end a response during thinking. Compare complete answers and total token cost.

  3. [ 03 ]

    Preserve conversation history

    Keep signed thinking blocks unchanged and history append-only. Native thinking replay is account-bound.

Route

One route, standard clients.

Authenticate with a MiniRouter bearer key. Caller-provided upstream credentials are not used on this route.
Read model setup docs
Base URLhttps://api.minirouter.sh/v1
Model IDanthropic/claude-haiku-5.5
Endpoints/v1/chat/completions/v1/messages/v1/responses
Parametersmax_tokens, stop, tools, tool_choice, reasoning, include_reasoning
Structured outputNot enforced; a requested format is dropped and the answer is plain text.
ModalitiesInput: text, image · Output: text

Configured providers. Priority follows the live catalog order.

ProviderPriorityInput / 1MOutput / 1MParametersAccess
vercel
1$0.105$0.525max_tokens, stop, tools, tool_choice, reasoning, include_reasoningMiniRouter balance

Every model, one key.

Get an API keyRead the docs