Skip to content

Claude Haiku 5.5 vs Claude Haiku 4.5

Claude Haiku 5.5 is from Anthropic, released October 7, 2026. Claude Haiku 4.5 is from Anthropic, released October 15, 2025.

  • LMArena Elo1425 vs 1414. Claude Haiku 5.5 is 0.8% ahead on LMArena Elo.
  • Input per 1M$0.10 vs $1. Claude Haiku 5.5 costs 90% less per million input tokens.
  • Output per 1M$0.50 vs $5. Claude Haiku 5.5 costs 90% less per million output tokens.
  • Context window1M vs 200k. Claude Haiku 5.5 takes 400% more context in one request.

Prices are the vendors’ published API rates; Elo is LMArena’s text board. Each figure is dated on its model page.

Interactive Comparison

Loading chart...
$0.10 / $0.50 per million tokens
$1 / $5 per million tokens
Identity
DeveloperAnthropicAnthropic
ReleasedOct 2026Oct 2025
Status—Available (legacy)
LicenceProprietaryProprietary
Self-hostableNoNo
Cost
Blended $/1M tokensinput × 0.75 + output × 0.25$0.200 / 1M tokens$2.000 / 1M tokens
Input price$0.100 / 1M tokens$1.000 / 1M tokens
Output price$0.500 / 1M tokens$5.000 / 1M tokens
Cached input$0.01 per 1M tokens (over 100K: $0.05)$0.10 / 1M tokens (cache hit)
Batch discount—50% off input/output (Batch API)
Free tier—Free access on claude.ai; small trial API credit for new accounts
Capacity
Context window1M tokens200k tokens
Max output128K tokens64k tokens
Long-context surchargePrompts over 100K tokens: $0.50 / $2.50 per 1M tokensN/A (200k context, not the 1M tier)
Capability
Vision inYesYes
Audio in—No
Function calling—Yes
Structured output—Yes
Extended reasoning—Yes
Web search—Yes
Code execution—Yes
Access
APIYesYes
Chat app—Yes
Cloud marketplaces—AWS Bedrock, Google Cloud, Microsoft Foundry
Fine-tuning—No
Measured quality
LMArena Elo1425 (checked Oct 2026)1414 (checked Oct 2026)
LMArena rankRank 126Rank 144
GPQA Diamond90 · max effort—
Elo per dollarLMArena Elo ÷ blended $/1M tokens7125707

Add a third model to this comparison