Claude Haiku 4.5
AnthropicOct 15, 2025N/A
Anthropic's fast small model. At launch it shipped near-frontier coding ability (~73% SWE-bench Verified) at roughly a fifth of Sonnet pricing — an aging but dependable budget pick now joined by cheaper rivals below.
- Status
- Generally available
- License
- Proprietary API
- Context window
- 200K
- Max output
- 32K
- Input · $/M
- $1
- Output · $/M
- $5
- Cached input · $/M
- $0.1
- Blended · $/M
- $2.00
- Parameters
- Undisclosed
- Inputs
- Text + images
Benchmarks
Composite 73.3· mean of published scores
Reasoning & knowledge
GPQA Diamond
no public score
MMLU-Pro
no public score
Humanity's Last Exam
no public score
ARC-AGI-2
no public score
Coding
SWE-bench Verified
73.3%
dataset best 93.4SWE-bench Pro
no public score
LiveCodeBench
no public score
Agents & computer use
Terminal-Bench 2
no public score
OSWorld 2.0
no public score
BrowseComp
no public score
Math
AIME 2026
no public score
Known strengths
- Very low latency
- ~73% SWE-bench V at small-model pricing