Zumik
Sign inSign up

Claude Haiku 4.5

The fastest, cheapest Claude - tool-heavy agent loops at a fraction of frontier cost, with explicit caching to keep stable prefixes cheap. The usual code.fast target under Anthropic policy.

Call it through ZumikAbout Anthropic
$1.00
Input / 1M tokens
$5.00
Output / 1M tokens
$0.10
Cache read · −90%
200K
Context window

At a glance.

ProviderAnthropic
Familyclaude-haiku
Released2025-10
LicenseProprietary
Context window200K tokens
Max output64K tokens
Modalitiestext, image, pdf
Tool callingYes
Reasoning modeYes
Cachingexplicit
Batch discount50% off

What reuse looks like here.

Claude Haiku 4.5 · agent trafficper request
Total input100%
Candidate reuse61%
Realized reuse55%
Capture rate90%
110ms
Warm TTFT · −66% vs cold
230
Output tokens / sec
Reuse economics

What you actually pay once caching works.

At a typical 55% prefix reuse, a million input tokens on Claude Haiku 4.5 effectively costs $0.51 instead of $1.00 - blending to roughly $1.63 with a 25% output share. Background work drops a further 50% on the batch tier.

Estimate it for your workload
Best for
fast-agentstool-callshigh-volume

Routes through these aliases:

code.fastauto.fastauto.cheapest

Same OpenAI client, this model.

python
from openai import OpenAI

client = OpenAI(base_url="https://api.zumik.ai/v1", api_key="zk_live_...")

r = client.responses.create(
    model="claude-haiku-4-5-20251001",          # or an alias like code.fast
    input="Draft a fix for the failing test.",
)
print(r.usage.input_tokens_cached)   # confirm reuse

How Claude Haiku 4.5 stacks up.

Head-to-head

Claude Haiku 4.5 vs Gemini 3.5 Flash

The fastest profiled models for latency-critical agent loops.

Other options for these workloads.

OpenAI gpt-oss-120b

$0.21 reuse-adj

GPT-5 Mini

$0.59 reuse-adj

Gemini 3.5 Flash

$2.82 reuse-adj

Claude Haiku 4.5, answered.

How much does Claude Haiku 4.5 cost?

Claude Haiku 4.5 is $1.00 per million input tokens and $5.00 per million output tokens through Zumik. Cache reads are $0.10 per million, a 90% discount on input.

What is Claude Haiku 4.5's context window?

Claude Haiku 4.5 supports a 200K-token context window with up to 64K output tokens.

Does Claude Haiku 4.5 support prompt caching?

Yes. Anthropic uses Explicit cache_control breakpoints caching. In the Zumik corpus, Claude Haiku 4.5 shows a median cache capture of 90% on agent workloads.

Which Zumik aliases route to Claude Haiku 4.5?

Claude Haiku 4.5 is a candidate for the code.fast, auto.fast, auto.cheapest aliases, selected when it wins under current routing policy.

Run Claude Haiku 4.5 with reuse measured.

Point an OpenAI client at Zumik and see exactly how much of this model's input you are reusing.

Migration quickstartBack to catalog