# Claude Haiku 218 Hacker News commenters recommended Claude Haiku and 134 criticized it, November 2025 to November 2026. It was recommended in 13 of 13 months; owners who said how long report a median of 0.3 years (4 owners). **Verdict:** A fast, cheap Anthropic model that handles well-defined jobs like JSON extraction, summaries and subagent work, though it struggles with real coding and rivals often cost less. ## Why people recommend it - Fast and cheap (8 comments): "Haiku is fastest and also quite good at this! Because the edits are quite accurate in terms of its this element with this value" ([source](https://news.ycombinator.com/item?id=47728817)) "just seeing the difference between Haiku 4.5 (a year ago) and Haiku 5.5 (today - and 1/10th the cost) was worth it alone" ([source](https://news.ycombinator.com/item?id=50001721)) "Haiku sits at this weird middle point where, if you have a well defined task, you can use a smaller/faster/cheaper model" ([source](https://news.ycombinator.com/item?id=46038058)) - Great for structured extraction (4 comments): "It continues to be absurdly capable for converting natural language to structured JSON based on a set of fairly complex business rules." ([source](https://news.ycombinator.com/item?id=49804734)) "Haiku 4.5 detected some footers that were on every page and moved them to be the footer at the end of the entire document" ([source](https://news.ycombinator.com/item?id=47431708)) "Even small, cheap models like Gemini Flash or Claude Haiku do an excellent job extracting pure content." ([source](https://news.ycombinator.com/item?id=46636252)) - Good cheap subagent (3 comments): "A subagent with a cheap model like haiku or similar is the way to go instead of dumping tool output directly into the main agent context" ([source](https://news.ycombinator.com/item?id=49663943)) "Mostly used Claude Haiku 4.5 like 75% of the time, where it failed, switched to the sonnet 4.5" ([source](https://news.ycombinator.com/item?id=45871582)) "Opus directs and starts haiku/sonnet subagents. Much more efficient than opus reading it all." ([source](https://news.ycombinator.com/item?id=49903599)) - Handles simple well-defined tasks (5 comments): "I used it on an agentic reasoning task recently (reading a chunk of information and drawing a written conclusion, not writing code) and it did just fine." ([source](https://news.ycombinator.com/item?id=48377363)) "I think Haiku is fine (e.g.) for any task that you could almost, but not quite, complete with (regex?) find and replace." ([source](https://news.ycombinator.com/item?id=47860733)) "To ask for web search, simple commands and research, or passing logs to analyze it's more than enough" ([source](https://news.ycombinator.com/item?id=49673525)) ## What people warn about - Weak at complex coding (3 comments): "Then Haiku doesn't work (unless your "complex feature" is just an api endpoint and some UI that consumes it)." ([source](https://news.ycombinator.com/item?id=47857014)) "I don't find Haiku useful for actual coding tasks. Good to, ehm, generate commit messages and summaries." ([source](https://news.ycombinator.com/item?id=48377476)) "I compare it to the Haiku models, where you have to go in and be surgical about your changes" ([source](https://news.ycombinator.com/item?id=48552621)) - Pricier than competing models (5 comments): "Claude-haiku-4.5 was not far behind and was much faster due to the absence of thinking tokens, but much more expensive." ([source](https://news.ycombinator.com/item?id=45877576)) "It was already cheaper and more powerful than Haiku 5.5 which was released like a full month after it" ([source](https://news.ycombinator.com/item?id=50133098)) "I don't see a reason to use Sonnet, Haiku, or the smaller GPT models, because their API rates are much higher" ([source](https://news.ycombinator.com/item?id=48868036)) - Confidently wrong answers (3 comments): "It fell short answering some of the questions that required some deeper statistical analysis" ([source](https://news.ycombinator.com/item?id=49999726)) "I know that figure is too low but believed it anyway. I've just asked for better figures and it's completely unable to answer accurately." ([source](https://news.ycombinator.com/item?id=49413462)) "they're blatantly confidently wrong often enough that the wasted time is a significant net negative for me" ([source](https://news.ycombinator.com/item?id=48377403)) ## Alternatives hackers recommend - [OpenAI Luna](https://hackerapproved.com/p/luna.md): 218 recommend - [Apple Foundation Models](https://hackerapproved.com/p/apple-foundation-models.md): 45 recommend - [Ministral](https://hackerapproved.com/p/ministral.md): 18 recommend - [Microsoft Phi](https://hackerapproved.com/p/phi.md): 16 recommend - [LFM2.5 2.6B](https://hackerapproved.com/p/lfm2-5-2-6b.md): 14 recommend ## Where to get it - [Visit the official site](https://www.anthropic.com/) ## Cite as Hacker Approved, "Claude Haiku", https://hackerapproved.com/p/claude-haiku/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).