haiku 5.5 is definitely a big jump over 4.5 in intelligence and efficiency, but i'd still pick luna high for everyday workloads
the capability-to-cost balance is hard to beat even though gpt-6 sol is a regression from 5.6 luna
makes me wonder why openai hasn't released 6.1 luna yet
Haiku 5.5 has 128k max output tokens and a 1M-token context window. It was run with default provider settings: temperature 1.0, default top-p and top-k, and with max reasoning effort.
Congrats to @AnthropicAI .
Full results at
Haiku 5.5 on Browser Use Bench v2.1 👀
> same performace as Luna
> 2.8x more expensive
Haiku gets 5x more expensive after 100k context, including cached tokens. BU Bench 2.1 is mostly extremely long-running horizon tasks. It hits that pricing tier a lot.
Haiku 5.5 Is A Cheap Model That Underperforms In Its Class
- MUCH worse than DeepSeek
- worse than Luna as well
Anthropic is better off launching Fable 5.5 🤷♀️
Haiku is 100x cheaper than Fable. 50x cheaper than Opus. And ~10x cheaper than GLM, its most popular open source peer. Great for users, bad for Anthropic (and tbf, OpenAI) revenue + valuations. Will see this tier grow among non-coding enterprise users...until hardware bottlenecks clear up and free local models devour it all
Claude Haiku 5.5 is now available in Glean. 🚀
It joins our lineup of efficient models for everyday tasks, with strong performance in our internal evals for finding information, synthesizing across sources, and generating structured output.
More model choice, so teams can use the right model for the work at hand.