Claude Haiku 4.5 — Spezifikationen
Compiled by Mustafa Ihsan from the vendor’s published documentation · Last updated
| Entwickler | Anthropic |
|---|---|
| Typ | LLM |
| Modality | Text, Vision → Text |
| Parameter | Nicht offengelegt |
| Kontextfenster | 200 K |
| Maximale Ausgabe | 64 K |
| Lizenz | Proprietär |
| Offene Gewichte | Nein |
| Veröffentlicht | 2025 |
| Eingabepreis | $1.00 /1M |
| Ausgabepreis | $5.00 /1M |
| API-Anbieter | Anthropic, AWS, Vertex AI, Azure |
What is Claude Haiku 4.5?
Claude Haiku 4.5 is Anthropic’s fast, low-cost tier: $1 per million input tokens and $5
per million output, with a 200K context window and 64K maximum output. It is built for the
high-volume, latency-sensitive end of a workload — classification, extraction, routing,
short-form generation, and the thousands of small calls an application makes that never
needed a frontier model in the first place.
Haiku’s real job in most production systems is to absorb traffic. A blended cost of about
$1.80 per million tokens means it can handle an order of magnitude more requests than an
Opus-tier model for the same money, which is why the standard architecture is Haiku in
front, escalating to Sonnet or Opus only when a confidence check fails. The 200K context is
the one place it is genuinely narrower than its siblings — the Sonnet and Opus tiers reach
1M — so document-scale work that needs the whole corpus in one prompt belongs elsewhere.
For everything that fits, Haiku 4.5 is usually the correct default, and treating it as the
baseline rather than the fallback is what keeps an AI feature’s unit economics viable.
Claude Haiku 4.5 pricing: API cost per 1M tokens
| Eingabe (pro 1 Mio. Token) | $1.00 |
|---|---|
| Ausgabe (pro 1 Mio. Token) | $5.00 |
| Verhältnis Output/Input | 5× |
| Gemischt (4:1 Input:Output) | $1.80 pro 1 Mio. Tokens |
What Claude Haiku 4.5 costs per month
Tatsächliche monatliche Ausgaben bei einem 4:1-Input-zu-Output-Mix – dem Verhältnis, das typische Chat- oder RAG-Arbeitslasten tatsächlich erzeugen.
| Workload | Tokens/Monat | Kosten pro Monat |
|---|---|---|
| Nebenprojekt | 1 Mio. Eingabe / 0,25 Mio. Ausgabe | $2.25 |
| Kleines Team | 20 Mio. Eingabe / 5 Mio. Ausgabe | $45 |
| Produktion | 200 Mio. Eingabe / 50 Mio. Ausgabe | $450 |
Stellen Sie Ihre eigenen Berechnungen im KI-API-Kostenrechner.
Cheaper alternatives to Claude Haiku 4.5
| Modell | Gewichteter Preis pro Million US-Dollar | Sie sparen |
|---|---|---|
| DeepSeek V4-Pro offenere | $0.522 | 71 % günstiger |
| Kimi K2.7 Code offenere | $0.980 | 46 % günstiger |
| DeepSeek V4-Flash offenere | $0.168 | 91% cheaper |
Häufig gestellte Fragen
How much does Claude Haiku 4.5 cost per 1M tokens?
Claude Haiku 4.5 costs $1.00 per 1M input tokens and $5.00 per 1M output tokens. At a typical 4:1 input-to-output mix that blends to about $1.80 per 1M tokens.
How much does Claude Haiku 4.5 cost per month?
A small-team workload of 20M input and 5M output tokens a month costs about $45 on Claude Haiku 4.5. A side project (1M in / 0.25M out) costs roughly $2.25.
What is a cheaper alternative to Claude Haiku 4.5?
DeepSeek V4-Pro is the strongest cheaper option in our database at $0.522 per 1M blended — about 71% less than Claude Haiku 4.5. It is also open-weight, so self-hosting is an option.
Can I run Claude Haiku 4.5 locally?
No. Claude Haiku 4.5 is a closed, API-only model — the weights are not released, so it cannot be self-hosted.
Why does Claude Haiku 4.5 charge more for output than input?
Output tokens are generated one at a time and cannot be batched the way a prompt can, so they cost the provider more to serve. Claude Haiku 4.5 charges 5× more for output, which is why prompt-heavy workloads are far cheaper to run than generation-heavy ones.
Alle Claude-Modelle im direkten Preisvergleich: Claude-API-Preise.
Die Preise entsprechen den offiziell veröffentlichten Listenpreisen für die primäre API des jeweiligen Modells und werden regelmäßig aktualisiert, sobald Anbieter diese ändern. Volumen-, Batch- und Cached-Input-Rabatte sind nicht enthalten. Vergleichen Sie alle Modelle nebeneinander im Datenbank für KI-Modelle oder das LLM-Leaderboard.