Anthropicproprietary
Released: 2024-10-22

Claude 3.5 Haiku

Anthropic rapid lightweight foundation model offering Claude 3 Opus-level performance at compact speed and cost.

Intelligence Index
76.5/ 100
Calibrated multi-domain composite
Throughput Speed
95tok/sec
Output streaming throughput
Time To First Token
310ms
Initial chunk server latency
LiveBench Contam-Free
—%
September 2026 suite score

Sourced Benchmark Evaluations (2)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
ARC-AGI-2Nearing Saturation58.5%
75
2024-11-19independentSource ↗
SWE-bench VerifiedSaturated48.6%
25
2024-10-31independentSource ↗
Anthropic Python SDKAPI Ready
import anthropic

client = anthropic.Anthropic()
message = client.messages.create(
    model="claude-3-5-haiku",
    max_tokens=2048,
    temperature=0,
    messages=[{"role": "user", "content": "Execute SWE-bench verified repository fix"}]
)
print(message.content[0].text)

Evaluation & Sourcing Notes

Scores listed for Claude 3.5 Haiku represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.