# Claude 3 Haiku ![Organization](https://img.shields.io/badge/Org-Anthropic-d97757) ![Context](https://img.shields.io/badge/Context-200K%20tokens-1f6feb) ![Pricing](https://img.shields.io/badge/Price-%240.25%20%2F%20%241.25%20per%201M-2da44e) ![License](https://img.shields.io/badge/License-Proprietary-d29922) ![Modality](https://img.shields.io/badge/Modality-Text%20%2B%20Vision-8957e5) ![Released](https://img.shields.io/badge/Released-Mar%202024-555) > **Quick answer:** Claude 3 Haiku is Anthropic's fastest and most affordable Claude 3 model, released March 13, 2024. It scores 20.2% on LiveCodeBench, 75.2% on MMLU, and 8.00/10 on MT-Bench. With a 200K-token context window at $0.25/$1.25 per 1M tokens, it was the leading cost-efficient model in the Claude 3 family at launch. ## At a Glance **Where Claude 3 Haiku leads** - Fastest Claude 3 model — lowest latency in the family - $0.25/$1.25 per 1M tokens — highly cost-efficient - 200K-token context window - Vision capability at the mini price tier - 8.00/10 MT-Bench — solid instruction following **Where it lags** - 20.2% LiveCodeBench — below Claude Haiku 3.5 and current frontier models - 75.2% MMLU — lower than Claude Sonnet 3.x and Opus - Superseded by Claude Haiku 3.5 ($0.80/$4) for better performance - August 2023 knowledge cutoff — significantly outdated **Best for:** Legacy integrations validated on Claude 3 Haiku; high-volume classification and summarisation tasks where maximum throughput at minimal cost is the priority. ## What Claude 3 Haiku Is Claude 3 Haiku is the small-model tier of Anthropic's Claude 3 family, released March 13, 2024 — the same day as the full Claude 3 family launch. It introduced vision capability at a cost point below all previous Claude models, targeting high-throughput production use cases. The model was designed around near-instant responsiveness: at launch it was Anthropic's fastest model, suited for real-time user-facing applications and moderation pipelines where latency matters more than peak accuracy. Claude Haiku 3.5 replaced Claude 3 Haiku as the recommended small-tier model in October 2024, offering substantially better performance ($0.80/$4 per 1M — more expensive but much stronger). For new deployments, Claude Haiku 3.5 is the correct choice. Claude 3 Haiku remains available for existing integrations. ## Specifications | Field | Value | |---|---| | Organization | Anthropic | | Parameters | Undisclosed | | Context window | 200,000 tokens | | Max output | 4,096 tokens | | License | Proprietary (API only) | | Release date | March 13, 2024 | | Knowledge cutoff | August 2023 | | Modality | Text + Vision (multimodal) | ## Pricing | | Input (per 1M tokens) | Output (per 1M tokens) | |---|---|---| | Anthropic API | $0.25 | $1.25 | Prompt caching available. Pricing per [Anthropic pricing page](https://www.anthropic.com/pricing). ## Context Window Claude 3 Haiku has a **200,000-token context window** — roughly **150 pages** of text in a single request. ## Public Benchmark Scores | Benchmark | Score | Source | Date | |---|---|---|---| | [LiveCodeBench](/benchmarks/livecodebench) | 20.2% | Benchgen evaluation | 2024 | | [MMLU](/benchmarks/mmlu) | 75.2% | Benchgen evaluation | 2024 | | [GSM8K](/benchmarks/gsm8k) | 88.9% | Benchgen evaluation | 2024 | | [MT-Bench](/benchmarks/mt-bench) | 8.00 / 10 | Benchgen evaluation | 2024 | ## Claude 3 Haiku vs Alternatives | Model | Context | MMLU | LiveCodeBench | Price (in/out per 1M) | |---|---|---|---|---| | **Claude 3 Haiku** | 200K | 75.2% | 20.2% | $0.25 / $1.25 | | [Claude Haiku 3.5](/models/anthropic/claude-haiku-3-5) | 200K | — | — | $0.80 / $4 | | [GPT-4o mini](/models/openai/gpt-4o-mini) | 128K | — | 27.5% | $0.15 / $0.60 | | [Gemini 2.5 Flash](/models/google/gemini-2-5-flash) | 1M | — | 61.9% | $0.15 / $0.60 | For new deployments, Claude Haiku 3.5 provides significantly better performance. GPT-4o mini and Gemini 2.5 Flash offer better price/performance in their respective tiers. ## Use Claude 3 Haiku via API ```python import anthropic client = anthropic.Anthropic(api_key="YOUR_API_KEY") message = client.messages.create( model="claude-3-haiku-20240307", max_tokens=1024, messages=[{"role": "user", "content": "Classify this support ticket..."}], ) print(message.content) ``` ## Related Models & Benchmarks - Compare: [Claude Haiku 3.5](/models/anthropic/claude-haiku-3-5) · [Claude Sonnet 3.x](/models/anthropic/claude-sonnet-3-7) · [GPT-4o mini](/models/openai/gpt-4o-mini) - Benchmarks: [LiveCodeBench](/benchmarks/livecodebench) · [MMLU](/benchmarks/mmlu) · [MT-Bench](/benchmarks/mt-bench) ## Frequently Asked Questions

What is Claude 3 Haiku?

Claude 3 Haiku is Anthropic's fastest Claude 3 model, released March 2024. It supports 200K-token context and vision input at $0.25/$1.25 per 1M tokens. It has been superseded by Claude Haiku 3.5 for new deployments.

What is Claude 3 Haiku's context window?

Claude 3 Haiku supports a 200,000-token context window with up to 4,096 output tokens.

How much does Claude 3 Haiku cost?

Claude 3 Haiku costs $0.25 per 1M input tokens and $1.25 per 1M output tokens. Claude Haiku 3.5 provides significantly better performance at $0.80/$4.

Should I use Claude 3 Haiku or Claude Haiku 3.5?

Claude Haiku 3.5 for new deployments — substantially better benchmark scores with a more recent knowledge cutoff, at 3× higher cost. Claude 3 Haiku is appropriate only for existing integrations where the API version (claude-3-haiku-20240307) is locked.

--- *Specs and scores sourced from Anthropic's Claude 3 announcement (March 2024) and Benchgen evaluations. Pricing cited to the Anthropic pricing page. Last updated 2026-07-24.*

Claude 3 Haiku — Training jobs

Training jobs