OpenRating is an independent, open-source AI rating platform. Join the community

OOpenRating
All models

Anthropic · Released Jan 18, 2026

Claude 4.5 Haiku

Anthropic's lightweight reasoning workhorse built for real-time subagents, interactive chat, and high-frequency classification.

Proprietary

OpenRating Index

49.6

0–100 composite

Reasoning Index

68.2

0–100 composite

Cost / Task

$0.22

benchmark run

Output speed

195 tok/s

blended tok/s

Time to first token

0.3s

median latency

Input price

$0.4

per 1M tokens

Output price

$1.6

per 1M tokens

Context window

200K

tokens

Strengths

  • 195 tok/s output speed with 0.35s TTFT

  • Extremely low latency for conversational UI

  • Efficient subagent execution in multi-agent graphs

How we rate Claude 4.5 Haiku

OpenRating scores every model on the same axes. The OpenRating Index is a weighted composite of reasoning, knowledge, and agentic benchmarks, normalised to a 0–100 scale. Speed is blended output measured in tokens per second under standardised load, with time-to-first-token reported separately. Prices are list input and output prices per million tokens, and Task Cost measures end-to-end workload spending.

OpenRating Index

49.6

Reasoning Index

68.2

Token Pricing

Input per 1M tokens

$0.4

Output per 1M tokens

$1.6

Cost of 1M input + 1M output

$2

Task Cost Economics

Empirical cost breakdown for a standard OpenRating Index benchmark task ($0.22 total).

Reasoning (Chain of Thought)
$0.086
Answer generation
$0.033
Cache write overhead
$0.055
Cache hit queries
$0.038
Input payload
$0.007