Anthropic has released Claude Haiku 5.5. This release delivers the company's fastest and most affordable small model to date. Benchmark results show a major performance jump over its predecessor. Anthropic is also cutting prices for Sonnet 5.5 alongside this launch.
Haiku 5.5 is designed for high-volume, cost-sensitive tasks. These tasks include summarization, database queries, classification, and live customer support. On average, the model costs about 75 percent less than Haiku 4.5. For requests with prompts up to 100,000 tokens, prices drop by up to 90 percent. Anthropic notes that these shorter prompts account for roughly 90 percent of all previous Haiku requests. Prompts longer than 100,000 tokens cost five times as much.
Haiku 5.5 uses an updated tokenizer that consumes slightly more tokens per task than its predecessor. A similar tokenizer change on Opus 4.x models caused token usage to jump about 30 percent. Real-world savings are likely smaller than the per-token prices suggest. Anthropic points out this technical detail regarding the tokenizer updates.
Read nextGoogle Releases Nano Banana 2.1 Image Model With Lower PricesHaiku 5.5 shows large benchmark improvements
Haiku 5.5 scores 1,620 on the knowledge benchmark GDPval-AA v2.1, which is more than double the 735 scored by its predecessor. On Humanity's Last Exam, it hits 45.9 percent without tools and 57.4 percent with tools. The previous version managed 10.2 percent and 18.7 percent on those tests.
The biggest performance jump appears in computer use. Haiku 5.5 scores 72.4 percent on OSWorld-2.1, up from 15.7 percent. On the agentic coding benchmark Terminal-Bench 4.0, it reaches 39.2 percent while Haiku 4.5 scored zero. Anthropic compares the model against OpenAI's budget model GPT-6 Luna, stating that Haiku 5.5 leads across every tested category. Sonnet 5.5 reference scores show that Haiku 5.5 still falls well behind Anthropic's larger model.
Haiku 5.5 is the first Haiku-class model with adjustable reasoning levels. This feature lets users balance cost against quality. Anthropic states the model works best for narrowly scoped tasks like compaction, summarization, or sub-agent work. Sonnet 5.5 and Opus 5.5 remain the better picks for complex agentic coding.
Cybersecurity safeguards are tighter than on the predecessor. They allow a broader range of defensive tasks than Sonnet 5.5 partly because the model is less capable overall. Penetration testing stays blocked. Organizations with broader needs can apply for verification programs for life sciences and cybersecurity.
Haiku 5.5 is available now across all platforms. These platforms include Amazon Web Services, Google Cloud, and Microsoft Azure.
Anthropic cuts Sonnet costs and adds API credits
Anthropic is cutting cache read costs for Sonnet 5.5 by 50 percent, dropping from $0.20 to $0.10 per million tokens. The company states this should reduce costs for most agentic tasks by about 20 percent. The change follows the introduction of OpenAI's new GPT-6.1 series.
Anthropic is also rolling out monthly API credits. Max-5x subscribers get $100, Max-20x subscribers get $200, and Team subscribers receive up to $500 per month. Users can spend these credits to experiment with tools, apps, and agents through the API. The company is updating its Python and TypeScript SDKs to add beta support for computer use and browser use.
