Model
Model explorer

Claude Haiku 5.5

CLOSED
Anthropic · Claude 5.5 family · released Oct 7, 2026

Model id claude-haiku-5-5, released 2026-10-07. Anthropic calls it its 'cheapest, fastest, and most capable small model'. It is priced 90% below Haiku 4.5 for prompts up to 100K tokens ($0.10/$0.50) and 50% below for longer prompts ($0.50/$2.50). It has a 1M context, 128K max output, and is the first Haiku with an effort parameter (default medium). It uses the newer tokenizer (~30% more tokens than Haiku 4.5). The launch table compares it only with Haiku 4.5, GPT-6 Luna and Sonnet 5.5. Cyber safeguards are looser than Sonnet 5.5's but stricter than Haiku 4.5's.

ReasoningCodingVisionFunction callingTool useAgentic
2922.0
Elo · rank #10
Parameters
Undisclosed
Active params
—
Context
1M tokens
Architecture
Proprietary model; architecture and parameter count undisclosed; first Haiku with an adjustable effort setting, 1M-token context
License
Proprietary
Languages
—
API price (in/out)
$0.1 / $0.5
Modalities
text · vision
Benchmark results
Bar shows position within the tracked field; marker = field best
ChartographyVision46.4%#4
best: Gemini 4 Argon · 71.6%
FrontierCodeCoding46.4%#10
best: Claude Opus 5.5 · 54.6%
Humanity's Last ExamReasoning57.4%#5
best: Claude Opus 5.5 · 67.7%
Terminal-Bench 4.0Agents39.2%#9
best: Claude Sonnet 5.5 · 70.6%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
$0.1
Output / M tok
$0.5
API price $0.1/$0.5 · each benchmark row carries its own source badge (see methodology)