We match an open model and the right software to your tasks, languages, licence requirements, and selected Core or Pro server. You receive a documented local AI setup, not a random folder of downloads.
Each option sets a practical range for memory and response time. We choose the exact model within that range when preparing your system.
Curated compact local assistant stack for private drafting and Q&A
Private chat, email drafting, summaries, meeting notes and day-to-day assistance, scaled to the RAM you choose next.
Curated local knowledge stack with retrieval, longer context and workflow tools
Balanced local reasoning for document-heavy work, private search, research, coding support and office assistants.
Curated high-memory stack for coding agents, orchestration and heavier secure workloads
High-memory local intelligence for coding agents, larger private knowledge bases, orchestration and heavier secure workloads.
These model families help us set the current SelbsAI options. The exact model we install may change when a newer open release performs better in independent tests.
A compact baseline for fast local assistants, offline drafting and lighter retrieval on compatible 16–32 GB appliances.
A local step-up for multilingual work, long-context tasks and richer instruction-following on compatible 32 GB appliances.
A model family we track for responsive multimodal chat, coding assistants and agentic workflows because Gemma 4 combines long context, native function-calling support and multi-token prediction drafters.
One of the main higher-capability open families we watch for larger reasoning, coding and multilingual deployments on high-memory local devices.
We review the publisher, licence, file format, hardware compatibility, and update history before considering a model for your system.
Publisher history, release notes, model-card quality, community usage, and maintenance signals are reviewed before a model is treated as a provisioning candidate.
The configurator now captures whether the customer wants permissive-only, commercial-ready, or restricted-model avoidance before final model selection.
Where supported, selbsai prefers formats and runtimes with clearer supply-chain posture, including Safetensors, GGUF, MLX packages, and established local runtimes.
The selected model/runtime stack is checked against RAM, VRAM, thermal budget, storage, context length, and the customer's target workloads.
We do not promise that every configuration will always use the same named model. A useful local AI server should use the right open model for the job and improve as better models and software become available.
Model rankings change. We link to current independent comparisons and official model cards so the basis for selection remains inspectable.
Independent model pages with direct comparisons across intelligence, speed, price, context window and methodology notes.
A widely used open-model benchmark hub for comparing community and lab releases across standard eval suites.
Useful for broad human-preference comparisons and keeping an eye on how major open releases stack up in live arena-style evaluation.
For invoice, receipt and document-heavy presets we pair the language model with open OCR and document-understanding tooling rather than relying on the base LLM alone.
Search-heavy presets use additional embedding and reranking components so large local indexes stay usable at real-world scale.
Each package combines a suitable model with tools for document search, coding, writing, or office work.
Local help for private repositories.
Draft, rewrite, summarize, extract.
Inbox work without inbox exposure.
Turns reading piles into briefings.
Find clauses, risks, gaps, and dates.
Prepare better conversations faster.
Policies and evidence, searchable locally.
Operations support from local records.
Stock lists, reorder issues, and reports.
Ask your manuals, folders, and notes.
Each option sets a practical range for memory and response time. We choose the exact model within that range when preparing your system.