TLDRocket
Sign in

Claude Haiku 5.5: Anthropic Cuts Prices by Up to 90 Percent

Trending Topics Jakob Steinschaden ● Covered by 11 sources

Anthropic just made Claude Haiku 5.5 much cheaper, with API prices cut sharply. It now matches OpenAI’s small model on price, right before Anthropic’s planned IPO.

Based on reporting by Trending Topics, Jakob Steinschaden — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Anthropic has pushed the price war down into its smallest models. Claude Haiku 5.5 is now the company’s cheapest and fastest small model, and for most requests it costs a tenth of Haiku 4.5 while Anthropic says it can do much more.

Through the API, the new model is priced at $0.10 per million input tokens and $0.50 per million output tokens for requests up to 100,000 tokens. Above that limit, the rate rises to $0.50 for input and $2.50 for output per million tokens. Anthropic says about 90 percent of past Haiku requests fall under that ceiling. There’s another wrinkle too: a new tokenizer counts the same text as about 30 percent more tokens, so the company says the typical cost reduction works out to about 75 percent.

Haiku 5.5 supports up to one million tokens of context and up to 128,000 output tokens. It is also the first Haiku model with adjustable effort, with medium set as the default. Anthropic is offering it through the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry. And the safety rules are tighter than before, with penetration testing blocked by default.

On Anthropic’s own benchmarks, the jump looks dramatic. In OSWorld, where models operate a computer, Haiku 5.5 scores 72.4 percent, far above Haiku 4.5’s 15.7 percent and ahead of OpenAI’s GPT-6 Luna at 48.9 percent. On Terminal-Bench, it reaches 39.2 percent versus 16.4 percent for Luna, though that result comes only at maximum effort. At the default setting, Anthropic says it scores about 20 percent. Sonnet 5.5 still leads on every test, and none of these numbers have been independently verified.

Anthropic is also trimming cached input pricing for Sonnet 5.5, from $0.20 to $0.10 per million tokens. Subscribers to the Max and Team plans get monthly API credits too, ranging from $100 for Max 5x to $500 for a team. The launch lands just as the company is reportedly preparing an IPO for mid-November, which makes the timing feel less like a product note and more like a market message.

My take — AI-written commentary, not fact-checked reporting

This is the kind of pricing move that says the small-model market is getting brutal, not healthy. Anthropic is clearly betting that lower prices will buy enough usage to matter before the IPO crowd starts asking awkward questions. Nice model, better margins please.

Read more about this at: Trending Topics

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.