According to AlphaMoat, in the latest data month Base Compute (basecompute.co) received 0 monthly visits — ranked #0 of 17,424 AI websites with traffic data, and #0 of 2,225 in General AI.
What is Base Compute
Base Compute is an AI inference lab working on running models on-device. Its consumer-facing piece is a local app for chat, meetings and coding that sets itself up for the machine it is installed on, with no cloud and no accounts, plus BaseRT, an LLM runtime for Apple Silicon that the site benchmarks against llama.cpp and MLX on prefill and decode. For chip vendors, model makers and device makers it runs an optimisation stack covering model quantisation, graph rewrites, operator fusion and kernels, tuned per model and per target device. Enterprise deployments are offered on-premise, hybrid or air-gapped.
- 01BaseRT LLM runtime for Apple Silicon
- 02Local chat, meeting and coding apps
- 03Model quantisation and operator fusion
- 04Custom kernels and hardware profiling
- 05Day-0 support for new model releases
- 06On-premise and air-gapped enterprise deployment
The local app runs on hardware you already own, described as zero $ per token and free to try. Commercial work is optimisation engagements for chip, model and device makers.
Popular in General AI
An all-in-one AI assistant for answering questions, writing, generating images, and coding.
+ Compare
A Google AI assistant combining writing, planning, and brainstorming.
+ Compare
An Anthropic AI assistant that analyzes files, code, and helps with writing.
+ Compare
An AI model developer focused on general intelligence foundation models and technology
+ Compare
An AI assistant that chats, generates images, writes code and connects to the web in real time
+ Compare
Data month: — · Visit figures are third-party traffic estimates aggregated monthly — best for scale and trends. Multiple domains of the same product are merged. See methodology.