NOUMENOpen app
NOUMEN

Anthropic · Efficient

Claude Haiku 4.5

Added Oct 15, 2025

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Claude Haiku 4.5 matches Sonnet 4's coding quality while running about twice as fast at roughly one-third the price, making it ideal for low-latency chat assistants, support bots, and pair programming. Pair it with Sonnet 4.5 to plan complex work and fan out execution across parallel Haiku workers.

  • Tools
  • Vision
  • Structured output
  • PDF input
Input
Text, Images
Output
Text
Discovery tier
Efficient

Pricing

USD per 1M tokens
Input
$1.00
Output
$5.00
Cache read
$0.10

Reference prices as published by the model provider. API example

Specifications

Context window
200K
Max output
64K

Side by side

Claude Haiku 4.5 compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Claude Haiku 4.5AnthropicThis modelClaude Opus 5.5AnthropicRemove
ProviderAnthropicAnthropic
Discovery tierEfficientFrontier
AddedOct 15, 2025Sep 22, 2026
Limits
Context window200K1MBest
Max output64K128KBest
Average output—2K tokens
Pricing, USD per 1M tokens
Input$1.00Best$4.00
Output$5.00Best$20.00
Cache read$0.10Best$0.20
Cache write—$5.00
Performance
Throughput—257.3 tok/s
Time to first token—5.4 s
Request success (24h)100.0%Best99.8%
Benchmark indexes
Intelligence Index—53.6
Capabilities
ReasoningNoYes
ToolsYesYes
VisionYesYes
Structured outputYesYes
PDF inputYesYes
Reasoning effort—low, medium, high, xhigh, max

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Claude Haiku 4.5
  • Claude Opus 5.5

Intelligence Index

  • Claude Haiku 4.5—
  • Claude Opus 5.553.6
3.8Catalog models by score55.1

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 63.2%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 33.8%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 1,683 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 1,704 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 28.8%

  • AA-LCRv1.1

    Reasoning across long documents

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 82.7%

  • MLCR-AA

    Long-context reasoning over medical records

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 15.0%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 55.6%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 56.6%

  • SciCode

    Scientific computing in Python

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 60.4%

Legacy evaluations

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Claude Haiku 4.5: —
    • Claude Opus 5.5: 82.7%

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Live routing data

Performance and reliability

Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.

Prompt cache hits
83.2%
About $0.25 per 1M input tokens with caching
Request success (24h)
100.0%
Reliable

Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.

Hourly request success · last 72 hoursLowest hour 96.7%
  • Healthy (98% or more)
  • Degraded (90–98%)
  • Down (under 90%)
  • No requests

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-haiku-4.5",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="anthropic/claude-haiku-4.5",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Anthropic

Other Anthropic models, starting with the same family.