Claude Haiku 3.5
Fast Claude model for lightweight app flows and retrieval-heavy workloads.
Best for: CheapestCoverage: fullRelease: GASynced within 7d
Pricing: Official LiveLimits: Curated OfficialBenchmarks: Third-Party
Current API snapshot
Official pricing, context, and latency normalized into a single current view.
Input price$0.80
Output price$4
Context window200K
Latency640 ms
Benchmark snapshot
Curated quality and performance signals from the latest accepted benchmark source.
Intelligence77
Coding73
Throughput142 tok/s
Benchmark sourceArtificial Analysis
Rank persona scores
Same ranking formulas used across the public tables, exposed directly for this model.
Cheapest76.3
Fastest71.1
Smartest0.0
Best value34.6
Formula reference
Persona definitions are fixed and public. The score meanings are strict, not inferred.
CheapestLowest total token cost only.Score = normalized total token cost only, where total cost = input price + output price.
FastestLowest observed latency only.Score = normalized latency only.
SmartestHighest intelligence benchmark score only.Score = normalized intelligence benchmark only.
Best ValueThe most balanced tradeoff between cost, quality, and responsiveness.Score = 45% intelligence + 35% cost + 10% latency + 10% context.
CodingHighest coding benchmark score only.Score = normalized coding benchmark only.
Sources
Every displayed field is tied back to a collected source and timestamp.
pricingofficial · collected Mar 10, 2026https://docs.anthropic.com/en/docs/about-claude/pricing
limitsofficial · collected Mar 9, 2026https://docs.anthropic.com/
benchmarksbenchmark · collected Mar 9, 2026https://artificialanalysis.ai/
Cost Calculator
Estimate daily and monthly spend for Claude Haiku 3.5.
Operational notes
Pricing notes: Live sync from Anthropic official pricing. Comparison uses base input, cache hit, and output token rates from the official model pricing table.
Data coverage: full
Max output tokens: 8.2K
Rate limit notes: Usage tiers and burst limits vary by account.
Updated: Mar 10, 2026