LLM ReferenceLLM Reference

Claude Haiku 4.5

claude-haiku-4-5

Researched 26d ago
ProprietaryMultimodalCodingRAGAgentsLong contextVisionJSON / Tool use

Claude Haiku 4.5 is worth evaluating for coding, rag, and agents when its provider route and context window match the workload.

Decision context: Coding task fit, 7 tracked provider routes, and research from 2026-04-19.

Use it for

  • Teams evaluating coding, rag, and agents
  • Workloads that can use a 200k context window
  • Buyers comparing 4 tracked provider routes

Do not use it for

  • Workloads where another current model has stronger sourced task evidence

Cheapest output

$4.00

AWS Bedrock per 1M tokens

Provider routes

7

Tracked API hosts

Quality / dollar

Grade D

Ranked by benchmark score divided by cheapest output price

Freshness

2026-04-19

Researched 26d ago

fresh

Top use-case fit

Coding

Q/$ D

1 relevant benchmark in the decision map.

RAG

Included by capability and metadata signals in the decision map.

Agents

Q/$ C

2 relevant benchmarks in the decision map.

Provider price ladder

Compare all 7
ProviderInput / 1MOutput / 1MBatch in / outRoute
AWS Bedrock$0.800$4.00-
Serverless
GCP Vertex AI$0.800$4.00-
Serverless
Anthropic$1.00$5.00$0.500 / $2.50
Serverless
OpenRouter$1.00$5.00-
Serverless

Benchmark peer barsfor Coding

Migration checks

No linked migration route is available for this model yet.

About

Claude Haiku 4.5 available on AWS Bedrock

Claude Haiku 4.5 has a 200K-token context window.

Claude Haiku 4.5 input tokens at $0.8/1M, output at $4/1M.

Capabilities

VisionMultimodalFunction CallingTool UseStructured OutputsCode Execution

Benchmark Scores(3)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionSource
BFCL68.7v4https://gorilla.cs.berkeley.edu/leaderboard.html
SWE-bench Verified73.3SWE-bench Verifiedhttps://www.swebench.com/verified.html
MultiChallenge50.5MultiChallengehttps://labs.scale.com/leaderboard/multichallenge

Rankings

Specifications

Released2025-10-01
Context200k
ArchitectureDecoder Only
Knowledge cutoff2025-02
Specializationgeneral
LicenseProprietary
Trainingpretrained

Created by

Developing safe and ethical AI systems.

San Francisco, California, United States
Founded 2021
Website