Anthropic: Claude Haiku 4.5 vs Anthropic: Claude Opus 4.5

Anthropic: Claude Haiku 4.5 is cheaper than Anthropic: Claude Opus 4.5 by about 80% on a blended (75% input / 25% output) basis. Anthropic: Claude Haiku 4.5 charges $1.000000 / $5.000000 per 1M input/output tokens; Anthropic: Claude Opus 4.5 charges $5.000000 / $25.000000.

FieldAnthropic: Claude Haiku 4.5Anthropic: Claude Opus 4.5
Input $/Mtok$1.000000$5.000000
Output $/Mtok$5.000000$25.000000
Cache read $/Mtok$0.100000$0.500000
Cache write $/Mtok$1.250000$6.250000
Input $/Mcharn/an/a
Output $/Mcharn/an/a
Context window200000200000
Tokenizerclaudeclaude

Price history

Same workload, monthly cost

Illustrative monthly cost at 10M tokens/month, split per workload preset. Not a substitute for the calculator, which uses your own volume, cache and batch assumptions.

WorkloadAnthropic: Claude Haiku 4.5Anthropic: Claude Opus 4.5
Coding agent$12.00$60.00
RAG chat$16.00$80.00
Summarisation (batch)$14.00$70.00
Classification$10.80$54.00
Chat assistant$26.00$130.00
Content generation$42.00$210.00

Cite this page

Anthropic: Claude Haiku 4.5 vs Anthropic: Claude Opus 4.5: $1.000000/$5.000000 vs $5.000000/$25.000000 per 1M tokens, as of 2026-10-04 (Token Barometer, https://tokenbarometer.com/compare/anthropic_claude-haiku-4-5-vs-anthropic_claude-opus-4-5).

Anthropic: Claude Haiku 4.5 page | Anthropic: Claude Opus 4.5 page | Open in calculator