Skip to content
llmwaves

Claude Haiku 4.5

Anthropic · released Oct 15, 2025

ProprietaryReasoningTool useStructured outputVision

Claude Haiku 4.5 is Anthropic’s fastest and most efficient model, delivering near-frontier intelligence at a fraction of the cost and latency of larger Claude models. Matching Claude Sonnet 4’s performance...

Specification

Context window
200K
Max output
64K
Knowledge cutoff
Not stated
Parameters
Undisclosed
Licence
Proprietary
Serving providers
4
Moderated
Yes
Uptime
100.0%

Intelligence

24.1

66th percentile

Coding

Coding Index

Agentic

16.5

Agentic Index

Output speed

111 t/s

Median across providers

Latency

487ms

Time to first token

Cost per task

$0.45

Estimated

Benchmarks

Where the score comes from

The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.

Evaluation scores

Percentage correct · higher is better

  • MMLU-Pro
    80.0%
  • GPQA Diamond
    64.6%
  • LiveCodeBench
    51.1%
  • AA-LCR (long context)
    48.3%
  • IFBench
    42.0%
  • AIME 2025
    39.0%
  • SciCode
    34.4%
  • τ²-bench (Telecom)
    32.5%
  • Terminal-Bench Hard
    27.3%
  • Humanity's Last Exam
    4.2%

An evaluation missing from this list was not run for this model — it is not a zero.

View as table
EvaluationScore
MMLU-Pro80.0%
GPQA Diamond64.6%
LiveCodeBench51.1%
AA-LCR (long context)48.3%
IFBench42.0%
AIME 202539.0%
SciCode34.4%
τ²-bench (Telecom)32.5%
Terminal-Bench Hard27.3%
Humanity's Last Exam4.2%

Against its peers

Intelligence Index · this model highlighted, nearest peers in grey

Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.

View as table
ModelIntelligence
Qwen3 Max24.5
Ling 3.0 Tiny (free)24.3
Claude Haiku 4.524.1
gpt-oss-120b24.1
Kimi K2 090524.0
o123.9
GLM 4.623.4
GLM 4.7 Flash23.3

Percentile among all indexed models

Intelligence66th
Agentic27th
Mathematics40th
Terminal work69th
Arena Elo35th

Pricing

What it costs to run

List prices per million tokens, plus what one representative task works out to.

List price

Input / 1M tokens
$1
Output / 1M tokens
$5
Cached input / 1M
$0.1
Blended 3:1
$2

One task, estimated

$0.45

Input tokens
50,000
Output tokens
80,000
Profile
Reasoning

Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).

Serving providers

Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.

Arena

Head-to-head generation quality

Elo from pairwise judgements, broken out by the kind of thing the model was asked to build.

Overall Elo
1,151
Win rate
45.6%
Strongest at
Data visualisation
Tournaments
1,131

Elo by category

Dot position on a 1,050–1,175 scale · Elo has no meaningful zero

Agent categories (full-stack apps, mobile apps) are only scored for models tested in agent harnesses.

View as table
CategoryEloWin rate
Data visualisation115145.6%
Websites114545.1%
Game development113844.6%
UI components113442.7%
3D scenes112741.1%
SVG107239.1%