Claude Haiku 4.5

AnthropicMULTIMODAL Proprietary
Provider
Anthropic
Modality
MULTIMODAL
Parameters
Context window
200,000 tokens
Weights
Proprietary
Released
15 Oct 2025
Compare Claude Haiku 4.5 with others

Claude Haiku 4.5 is the speed-and-cost tier — surprisingly capable coding (73% SWE-bench) at the lowest Claude price point.

Best for

  • Fastest, cheapest Claude tier for high-volume, low-latency work
  • Classification, extraction, routing, lightweight chat
  • Simple-to-moderate coding with strong value

How it compares — and the India angle

  • Smaller 200K context (vs 1M on Opus/Sonnet) — not for whole-repo or huge-document jobs.
  • For high-throughput Indian startups, Haiku's $1/$5 pricing keeps unit economics viable where Opus wouldn't.
  • Step up to Sonnet when a task needs deeper reasoning or the 1M window.

Benchmarks

Representative public scores (approximate, higher is better) for relative comparison. Check the provider for the latest official results.

SWE-bench Verified73

How to access

API ~$1 / 1M input, ~$5 / 1M output. The cheapest Claude tier.

Access Claude Haiku 4.5
Share: