--- title: I ditched Ollama as my default runtime, and the... description: A developer switched their default local LLM runtime from Ollama to BaseRT, a newer runtime built specifically for Apple Silicon using Metal directly rather... canonical: https://daily.dev/posts/i-ditched-ollama-as-my-default-runtime-and-the-replacement-starts-models-in-a-fraction-of-the-time-zsly2i005 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: I ditched Ollama as my default runtime, and the replacement starts models in a fraction of the time | daily.dev og:description: A developer switched their default local LLM runtime from Ollama to BaseRT, a newer runtime built specifically for Apple Silicon using Metal directly rather... og:url: https://daily.dev/posts/i-ditched-ollama-as-my-default-runtime-and-the-replacement-starts-models-in-a-fraction-of-the-time-zsly2i005 og:image: https://api.daily.dev/og/posts/zsLY2i005.png og:image:alt: I ditched Ollama as my default runtime, and the replacement starts models in a fraction of the time og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # I ditched Ollama as my default runtime, and the replacement starts models in a fraction of the time **[XDA Developers](https://daily.dev/sources/xda-developers)** · 5 min read · 0 upvotes · 0 comments ## Summary A developer switched their default local LLM runtime from Ollama to BaseRT, a newer runtime built specifically for Apple Silicon using Metal directly rather than MLX, PyTorch, or Core ML. On an M5 MacBook, BaseRT loads models much faster and improves prefill speed significantly (up to 6.4x faster than llama.cpp and 3.9x faster than MLX per BaseRT's own benchmarks), with more modest gains in token generation. Ollama remains more convenient and broadly compatible across platforms and hardware, and now has its own MLX backend in preview, but BaseRT's narrower focus on Apple Silicon yields better performance for supported models. Installation is a single curl script, and BaseRT exposes an OpenAI-compatible API, though model compatibility and format support still lag behind Ollama. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is BaseRT and how does it compare to Ollama for running local LLMs on a Mac? BaseRT is a local LLM runtime built specifically for Apple Silicon, using Apple's Metal API directly instead of relying on MLX, PyTorch, or Core ML. On an M5 Pro, its published benchmarks show up to 6.4x faster prefill than llama.cpp and 3.9x faster than MLX, plus faster model loading than Ollama, though it supports fewer models and platforms than Ollama. _Anyone weighing local LLM runtimes for Apple Silicon can track comparisons like this on daily.dev._ ### How do I install BaseRT on macOS? Install BaseRT on an Apple Silicon Mac running macOS 14 or later by running curl -LsSf https://basecompute.co/install.sh | sh in Terminal. The first model pull takes longer because BaseRT downloads and converts it into its own .base format, which supports quantization from Q2 to Q8; installed models can be listed with basert list. _Developers setting up local LLM tooling on Apple Silicon can find setup guides like this on daily.dev._ ### Why does my local LLM feel slow to respond after it has been idle for a while? Ollama unloads models from memory after a period of inactivity, so the next request has to reload the model before it can respond, adding a noticeable delay. Runtimes optimized for specific hardware, such as BaseRT on Apple Silicon, can load models in a fraction of that time, reducing the wait before the first token appears. _Developers chasing snappier local inference can follow runtime performance discussions on daily.dev._ ## Similar posts on daily.dev - [Your Local LLM Could Be Leaving Mac Performance on the Table](https://daily.dev/posts/your-local-llm-could-be-leaving-mac-performance-on-the-table-fxxfy93gy) · Medium · 0 upvotes · 0 comments - [I switched my local LLM setup to Ollama's new MLX engine, and my Mac suddenly feels twice as fast](https://daily.dev/posts/i-switched-my-local-llm-setup-to-ollama-s-new-mlx-engine-and-my-mac-suddenly-feels-twice-as-fast-3pyxiagbi) · XDA Developers · 1 upvotes · 0 comments - [Docker turned my AI setup into something I can rebuild in four minutes](https://daily.dev/posts/docker-turned-my-ai-setup-into-something-i-can-rebuild-in-four-minutes-rcpqm6blv) · XDA Developers · 0 upvotes · 0 comments --- Tags: [#mac](https://daily.dev/tags/mac), [#local-ai](https://daily.dev/tags/local-ai), [#ai-inference](https://daily.dev/tags/ai-inference), [#ollama](https://daily.dev/tags/ollama) [View this post on daily.dev](https://daily.dev/posts/i-ditched-ollama-as-my-default-runtime-and-the-replacement-starts-models-in-a-fraction-of-the-time-zsly2i005) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"I ditched Ollama as my default runtime, and the replacement starts models in a fraction of the time","url":"https://daily.dev/posts/i-ditched-ollama-as-my-default-runtime-and-the-replacement-starts-models-in-a-fraction-of-the-time-zsly2i005","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/i-ditched-ollama-as-my-default-runtime-and-the-replacement-starts-models-in-a-fraction-of-the-time-zsly2i005"},"datePublished":"2026-08-18T11:32:53.953Z","dateModified":"2026-08-18T11:43:29.101Z","description":"A developer switched their default local LLM runtime from Ollama to BaseRT, a newer runtime built specifically for Apple Silicon using Metal directly rather...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ff399a28072cbcfbc61407fc3256cf7b?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ff399a28072cbcfbc61407fc3256cf7b?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"XDA Developers","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"XDA Developers","logo":"https://media.daily.dev/image/upload/s--kCg6yyAP--/f_auto,q_auto/v1774964407/logos/xda-developers?_a=BAMAMiWQ0","url":"https://daily.dev/sources/xda-developers"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/i-ditched-ollama-as-my-default-runtime-and-the-replacement-starts-models-in-a-fraction-of-the-time-zsly2i005","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"mac,local-ai,ai-inference,ollama","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"XDA Developers","item":"https://daily.dev/sources/xda-developers"},{"@type":"ListItem","position":3,"name":"I ditched Ollama as my default runtime, and the replacement starts models in a fraction of the time"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/i-ditched-ollama-as-my-default-runtime-and-the-replacement-starts-models-in-a-fraction-of-the-time-zsly2i005#faq","mainEntity":[{"@type":"Question","name":"What is BaseRT and how does it compare to Ollama for running local LLMs on a Mac?","acceptedAnswer":{"@type":"Answer","text":"BaseRT is a local LLM runtime built specifically for Apple Silicon, using Apple's Metal API directly instead of relying on MLX, PyTorch, or Core ML. On an M5 Pro, its published benchmarks show up to 6.4x faster prefill than llama.cpp and 3.9x faster than MLX, plus faster model loading than Ollama, though it supports fewer models and platforms than Ollama. Anyone weighing local LLM runtimes for Apple Silicon can track comparisons like this on daily.dev."}},{"@type":"Question","name":"How do I install BaseRT on macOS?","acceptedAnswer":{"@type":"Answer","text":"Install BaseRT on an Apple Silicon Mac running macOS 14 or later by running curl -LsSf https://basecompute.co/install.sh | sh in Terminal. The first model pull takes longer because BaseRT downloads and converts it into its own .base format, which supports quantization from Q2 to Q8; installed models can be listed with basert list. Developers setting up local LLM tooling on Apple Silicon can find setup guides like this on daily.dev."}},{"@type":"Question","name":"Why does my local LLM feel slow to respond after it has been idle for a while?","acceptedAnswer":{"@type":"Answer","text":"Ollama unloads models from memory after a period of inactivity, so the next request has to reload the model before it can respond, adding a noticeable delay. Runtimes optimized for specific hardware, such as BaseRT on Apple Silicon, can load models in a fraction of that time, reducing the wait before the first token appears. Developers chasing snappier local inference can follow runtime performance discussions on daily.dev."}}]} ```