Anthropic Claude Haiku 4.5 by DigitalOcean

$10 off the fastest Kimi K3 and GLM 5.2$10 off the fastest Kimi K3 and GLM 5.2

Model details

Anthropic Claude Haiku 4.5

Claude Haiku 4.5 is Anthropic's fastest and most cost-efficient model, positioned as a lightweight variant that matches Sonnet 4's performance on coding, computer use, and agent-style tasks. Anthropic's announcement highlights a SWE-bench Verified score of 73.3%, placing the model among the strongest coding systems available at its size and price tier. Independent overviews echo this framing, describing Haiku 4.5 as delivering near-frontier capability at baseline cost, making it suitable for production workloads where latency and price matter as much as raw reasoning quality.

Practically, Haiku 4.5 is aimed at developers building agent pipelines, automated coding assistants, and high-volume customer-facing features that need multimodal input handling. Its pricing structure starts at roughly one dollar per million input tokens and five per million output tokens, with prompt caching offering up to ninety percent savings on cached reads. With a large context window suitable for retrieving over codebases or document collections, plus support for tool and function calling, it functions well as a sub-agent brain or a budget-friendly primary model for teams scaling inference across many requests.

DigitalOceananthropic-claude-haiku-4.5claude-haiku

Open provider page Provider docs

Quick Info

Provider: DigitalOcean
Model key: anthropic-claude-haiku-4.5
Release date: Oct 15, 2025
Last updated: Oct 15, 2025
Knowledge cutoff: 2025-02-28
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $1.00
Output token cost: $5.00

Limits

Output tokens: 8,192 tokens
Context window: 200,000 tokens

Latest news about Anthropic Claude Haiku 4.5

DigitalOcean

Official sourceAnnouncement

DigitalOcean published an architecture deep-dive on June 15, 2026 describing its Inference Router, built on the "Plano" control plane with Envoy, WASM filters, and async Rust components, and a ranking engine fed by live cost and latency data. Throughout the piece, Haiku is repeatedly used as the canonical cheap-model e The post argues against hardcoded intent-classifier routing in the application layer (calling it "double taxation") and instead positions DigitalOcean's router as a purpose-built, model-agnostic optimization layer that can route "easy" requests to Haiku-class models and keep frontier capacity for hard ones. For develop

DigitalOcean

Official sourcePreview

DigitalOcean announced Model Evaluations in Public Preview on its Inference Engine on June 4, 2026, and in its walkthrough scenario it explicitly lists Claude Haiku 4.5 as one of three tasks inside an Inference Router configuration alongside DeepSeek R1 Distill Llama 70B and Gemma 4, compared against Claude Sonnet 4.6 The post frames Model Evaluations as a reproducible workflow to compare routing strategies (always-frontier vs. task-specific fine-tune vs. Inference Router policy) on the same dataset, judge, and metrics, with the legal-assistant example designed to surface cost vs. p95 tradeoffs across long documents. For developers

DigitalOcean

CoverageBenchmark

Anthropic: Claude Haiku 4.5 by Anthropic. 200K context, from $1.00/1M tokens, vision, tool use, function calling. See benchmarks, comparisons, and our expert...

DigitalOcean

CoverageBenchmark

Claude Haiku 4.5 delivers near-frontier AI performance at baseline cost. Read on to find full benchmark analysis, feature breakdown, and how to get started.

DigitalOcean

CoverageComparison

GPT-5.4 Mini is cheaper and faster than Claude Haiku 4.5 with better benchmarks. Compare both models for sub-agent use cases and token efficiency.

DigitalOcean

CoverageComparison

Compare LLM inference API pricing across 17 providers. Find the best rates for Mistral Nemo, Qwen 3 235B, Llama 3.2 11B Vision from $0.10/M tokens. Compare GPT-4, Claude, Llama, and more.