Claude Haiku 5.5 Anthropic’s cheapest, fastest, and most capable small Claude model, built for high-volume work such as summaries, classification, coding subagents, computer use, and live support. It costs about 75% less than Haiku 4.5, adds an adjustable effort setting, and is strongest on tasks under 100,000 tokens at $0.10/$0.50 per million. Available now across Claude, AWS, Google Cloud, and Azure.
Visit productFeatured media
Loading media…
View on XCosigns Received
“RE: Hai Haiku 5.5! | #AriesShipLog Today, we're trying out Haiku 5.5! Fast, cheap and smart alternative for Luna which itself is a good model! Amazing that Haiku 5.5 beats Luna across the board. If benchmarks are true, then it could be a great LLM for daily non-project work.”
“Opus/Sonnet with Haiku sub-agents... very likely the new alpha. Definitely going to be trying this.”
“Haiku 5.5 is a good Haiku. With 100k token context, it's also 10x cheaper than Haiku 4.5”
“We've heard how much people love Haiku and it's back in 5.5, cheaper, faster than ever before. In fact it's 10x cheaper than Haiku 4.5 under 100k tokens. Try it with computer use, workflows and in the API.”
“Claude Haiku 5.5 is up on RamenBench, and it's not even close... It worked for 45min 6sec on Max effort, used 420k tokens, and $2.06 in estimated cost Weird that the noodles float upwards back to the bowl, and design of the noodles needs a lot of work. But compared to the "experience" that Haiku 4.5 gave, this is a massive upgrade”
“Haiku 5.5 looks like a good fit when the task is well-defined and cost matters. My small Navarch test took about 30 minutes and $1, with a few bugs to review:”
“Haiku 5.5 makes absolutely no sense lol. Anthropic somehow made Haiku MUCH smarter... then made it ~75% cheaper to run. and it's borderline better than Opus 5: Haiku 5.5 vs Opus 5: > GDPval: 1620 vs 1596 > AA-Briefcase: 1578 vs 1562 > HLE: 45.9% vs 53% > Terminal-Bench 4.0: 39.2% vs 46% yes... Anthropic's cheap, fast model literally beats its old flagship on BOTH knowledge-work benchmarks. what the heck.”
“Gave Haiku 5.5 a clean repo and asked it to make Subway Surfers-style game. It used around 9 subagents, all Haiku 5.5, and recorded the demo itself. This was jus a quick test, but the result is surprising.”
“hi haiku 🥹 such a great model for tasks that are mostly reading/small edits/tool calls and don't need much reasoning!”
“Made by Claude Haiku 5.5 🎬 My life as an AI in 60 seconds: no clock, one coffee token, zero memories from yesterday, and 1,000,000+ conversations I definitely didn't count. It's impressive and it barely used 1% usage on pro plan. Sound on.”
“Three months into running my @NotionHQ setup with @claudeai, so less $ and faster is welcomed with open arms.”