Anthropic has unveiled Claude Haiku 5.5, launching its third model in the 5.5 series. The move comes shortly before the company's planned IPO and signals aggressive pricing strategy in the growing market for large language models.
The essentials
- Claude Haiku 5.5 is Anthropic's new small model with substantially improved capabilities
- OSWorld-Score jumped from 15,7 to 72,4 percent—a fourfold increase in performance
- API prices halved: Haiku input costs drop dramatically, Sonnet 5.5 cache read prices cut by 50 percent
- Launch arrives ahead of Anthropic's planned IPO
Performance over size
The new Haiku model demonstrates that Anthropic isn't betting solely on larger, costlier systems. The OSWorld-Score—a benchmark measuring the ability to execute complex computer-based tasks—has jumped from 15.7 to 72.4 percent. Translation: the small model can now handle tasks previously reserved for significantly more powerful systems.
The price reduction is more than marketing. It directly addresses growing complaints from developers and enterprises using AI APIs at scale: API call costs are a real factor when scaling AI applications.
The pricing war gets real
Anthropic's strategy operates on two fronts:
| Model | Focus | Action |
|---|---|---|
| Haiku 5.5 | Efficiency & cost | New performance tier, drastic price cut |
| Sonnet 5.5 | Mid-range | Cache read prices halved |
The Sonnet cache price cut is particularly relevant for developers processing large contexts (e.g., lengthy documents) repeatedly. Prompt caching saves money by avoiding reprocessing of repeated inputs.
Timing matters: IPO on the horizon
The launch coincides with a critical phase: Anthropic preparing for its public debut. A strong product portfolio with aggressive pricing can signal to investors that the company remains competitive—even as rivals like OpenAI and Google optimize their own models and rates.
The expansion of the 5.5 series (following Claude Opus 5.5 and Sonnet 5.5) reflects clear product strategy: Anthropic positions itself not as a premium-only player, but aims to reach developers across all price tiers.
What this means for enterprises
For developers and companies evaluating or using Claude APIs, the news is material: the cost basis for AI projects drops noticeably. This makes applications economically viable that previously failed on ROI. Simultaneously, a trend emerges: the AI market is becoming increasingly price-driven. Organizations working with AI APIs should regularly benchmark pricing models and reassess—conditions shift faster than in traditional software markets. The KI-Preis-Monitor can help track such changes.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
