🌥 Claude Haiku 4.5: Light-Weight Model by Anthropic

Haiku 4.5, a new Anthropic's light-weight LLM is truly long-awaited: the startup released the last version of the compact model, Haiku 3.5, almost a year ago.

The model is more than twice faster than Sonnet 4, which makes Haiku 4.5 especially useful for agent tasks: computer use, quick coding, and creating fast prototypes. According to benchmarks, Haiku outperforms Sonnet 4 and only slightly falls behind Sonnet 4.5, GPT-5, and Gemini 2.5 Pro.

💰 At the same time, the model is three times cheaper than Sonnet, only $1/$5 per 1M input/output tokens. Anthropic's models are practically the best AI-coders, still Chinese competitors like GLM 4.6, Qwen3-Coder, and Kimi K2 offer similar performance at a price almost ten times lower than Sonnet.

➡️ You can try Haiku 4.5 for free here.

@hiaimediaen