← Models

Claude Haiku 4.5

Anthropic · Claude 4

Fast and efficient Claude tier for latency-sensitive assistant and automation workloads.

Part of Claude Haiku family
Type
multimodal
Context
200K tokens
Max Output
64K tokens
Status
current
Input
$1/1M tok
Output
$5/1M tok
API Access
Yes
License
proprietary
fast cost-efficient assistant automation enterprise
Released October 2025 · Updated July 6, 2026

Overview

Freshness note: Model capabilities, limits, and pricing can change quickly. This profile is a point-in-time snapshot last fully verified on June 8, 2026; comparison and routing references were refreshed on July 6, 2026.

Claude Haiku 4.5 is Anthropic’s current small-model route for high-throughput workloads. Anthropic’s current model page describes it as the fastest and most cost-efficient Claude model, useful for teams optimizing latency and cost while retaining stronger quality than bare-minimum baseline models.

Anthropic’s lifecycle page lists claude-haiku-4-5-20251001 as active with no tentative retirement sooner than October 15, 2026.

Capabilities

Anthropic explicitly positions Haiku 4.5 around speed, coding responsiveness, computer use, and parallelized subagent work. In practice, that makes it strongest in concise summarization, extraction, routing, lightweight coding, and customer-facing or internal operational flows that require quick response cycles.

Technical Details

Anthropic’s launch and current model page frame Haiku 4.5 as the throughput-oriented Claude where speed and predictable cost matter most. The current model overview lists:

  • Claude API model ID: claude-haiku-4-5-20251001
  • Claude API alias: claude-haiku-4-5
  • Context window: 200K tokens
  • Max output: 64K tokens
  • Extended thinking: supported
  • Adaptive thinking: not supported

The practical takeaway is not just “cheap assistant model.” It is a model that can handle coding, computer-use, and orchestration tasks at a price point that supports scaled deployment. It is still best used with explicit format constraints and robust post-validation for critical outputs.

Pricing & Access

Current Anthropic pricing for Claude Haiku 4.5 is 1per1Minputtokensand1 per 1M input tokens and 5 per 1M output tokens. Prompt caching is listed at 1.25per1Mtokensfor5−minutecachewrites,1.25 per 1M tokens for 5-minute cache writes, 2 for 1-hour cache writes, and 0.10forcachehitsandrefreshes;BatchAPIpricingis0.10 for cache hits and refreshes; Batch API pricing is 0.50 input and $2.50 output per 1M tokens. Access is available through Claude.ai on web, iOS, and Android, the Anthropic API, Amazon Bedrock, Google Cloud Vertex AI, Microsoft Foundry, Claude products, and Claude Code.

Best Use Cases

Best for support summarization, ticket routing, knowledge normalization, coding subagents, and lightweight coding or documentation workflows where cost and speed dominate.

Comparisons

Compared with Claude Sonnet 5, Haiku 4.5 offers better cost and latency with a lower ceiling on complex reasoning. Compared with GPT-5 nano, selection depends on ecosystem and output-style fit. Compared with Gemini 2.5 Flash-Lite, both target efficient production use with different platform tradeoffs.