Entirely local
Ollama or llama.cpp + Open WebUI + SQLite + locally stored documents.
No model API subscription; requires a capable computer and locally licensed model.
RESEARCH · OPEN TECHNOLOGY · PUBLIC-INTEREST AI
Developed as a cost-first reference for people building their own AI systems. The register distinguishes truly local software, limited free cloud access, commercially hosted services and existing Manifold Grace infrastructure. It supports the AI Public Adviser and the wider AI systems roadmap.
Audit status: 10 November 2026. 63 tool entries are documented; these are candidates, not installed products. No new accounts, API keys, deployments or external provider integrations have been created by this register. Free-tier usage limits and licences can change. Not all free software performs like paid frontier services, and computer hardware still costs money.
Ollama or llama.cpp + Open WebUI + SQLite + locally stored documents.
No model API subscription; requires a capable computer and locally licensed model.
Gemini or GroqCloud + Cloudflare Workers AI + Supabase, subject to current quotas.
Useful when local hardware is insufficient; credentials and privacy conditions require attention.
Existing GitHub Pages and GitHub Actions + permitted feeds + source verification + SQLite/Qdrant + model fallback.
Preserves a free public information layer and avoids premature dependency on one provider.
Expanded 2026-11-10: 36 further entries added to the original 27. New candidates include free hosted inference, local runtime alternatives, evidence ingestion, document parsing, workflow builders, evaluation, security, hosting, accessible speech and data management. No additional software has been installed or connected. Free tiers are limited and can change.
Read the overview first, then filter by technology category or candidate. The full table can be scrolled horizontally without truncating records. Each tool now includes a risk field; these are risk categories, not accusations or independently verified current weaknesses in any named organisation.
63 entries displayed
| No. | Tool and official link | Function | Availability | Account requirement | Free limits and real costs | Potential use | Next validation | Security, geopolitical and information risks |
|---|---|---|---|---|---|---|---|---|
| 1 | Ollama | Model runtime | Local open-source | No account required for local models | No recurring software licence; hardware costs | Local execution; simple model serving | Evaluate | Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 2 | llama.cpp | Model runtime | Local open-source | No | Free code; hardware costs | Quantised models, CPU/GPU and local API | Evaluate | Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 3 | Jan | Model runtime | Local open-source | No for local models | Free desktop software; hardware costs | Local AI desktop interface | Alternative | Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 4 | LM Studio | Model runtime | Free desktop software | No for local use | Free software, not all components open source | Local model exploration and server | Alternative | Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 5 | Open WebUI | AI interface | Free self-hosted edition | No hosted signup for self-host | Self-managed storage and maintenance | User interface, RAG and local/cloud connectors | Evaluate | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 6 | Google AI Studio / Gemini API | Hosted AI | Limited free API tier | Google account + API credential | Selected models free; usage limits and free-tier data use | Cloud reasoning benchmark or low-volume fallback | Evaluate | External processing, retention, lawful access and provider outages; verify controls before handling sensitive data |
| 7 | GroqCloud | Hosted AI | Limited free service | Account + API key | Requests/tokens per day vary by model | Fast inference alternatives | Evaluate | External processing, retention, lawful access and provider outages; verify controls before handling sensitive data |
| 8 | Cloudflare Workers AI | Hosted AI | Daily free allocation | Cloudflare account | 10,000 free neurons daily; supported models only | Small cloud inference and backend functions | Evaluate | External processing, retention, lawful access and provider outages; verify controls before handling sensitive data |
| 9 | Hugging Face | Hosted AI | Provider-dependent credit terms | Account | Credit availability and eligibility must be verified; not unlimited | Model discovery and trial provider access | Research only | External processing, retention, lawful access and provider outages; verify controls before handling sensitive data |
| 10 | LiteLLM | Model routing | Open-source library/gateway | No for local gateway | Third-party inference still billed | Provider-neutral switching, logging and limits | Evaluate | Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; Routing to alternate providers can change data jurisdiction and usage conditions; verify controls before handling sensitive data |
| 11 | Crawl4AI | Research retrieval | Apache-licensed self-host + hosted free tier | No locally; account/key for cloud | Local compute; hosted 1,000 monthly credits advertised | Extract accessible public sources for RAG | Evaluate | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; verify controls before handling sensitive data |
| 12 | GPT Researcher | Research agent | Open-source implementation | No for code; some providers require keys | Search and inference may cost | Research planning and multi-source synthesis | Evaluate | Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; verify controls before handling sensitive data |
| 13 | GDELT | Information feeds | Public research data | Varies by endpoint | Access policies, volume and licensing vary | Global news/event indicators | Evaluate | Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; verify controls before handling sensitive data |
| 14 | RSS and official feeds | Information feeds | Generally public feeds | Typically none for public RSS | Publisher rights and access rules vary | Low-cost dated primary-source ingest | Evaluate | Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; verify controls before handling sensitive data |
| 15 | Supabase | Structured database | Free hosted development tier | Account + project | 500MB database and other caps; inactivity pauses | Evidence register, audit trail and metadata | Evaluate | External processing, retention, lawful access and provider outages; Misconfigured access, backups, cross-user leakage and retention; verify controls before handling sensitive data |
| 16 | SQLite | Structured database | Public-domain library | No | Disk capacity and operational maintenance | Small durable local claim store | Evaluate | Misconfigured access, backups, cross-user leakage and retention; verify controls before handling sensitive data |
| 17 | Qdrant | Vector search | Free local/open-source or hosted trial cluster | No local; account for cloud | Hosted free 1GB RAM, 4GB disk; inactivity policy | Semantic source retrieval | Evaluate | External processing, retention, lawful access and provider outages; Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; Misconfigured access, backups, cross-user leakage and retention; verify controls before handling sensitive data |
| 18 | Chroma | Vector search | Open-source local | No | Local compute/storage | Local prototyping and retrieval | Alternative | Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; Misconfigured access, backups, cross-user leakage and retention; verify controls before handling sensitive data |
| 19 | pytest | Evaluation | Open-source | No | Compute and maintenance | Deterministic correctness tests | Available | Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 20 | Ragas | Evaluation | Open-source framework | No for code | Model evaluations may require paid inference | Evaluate RAG groundedness and answer quality | Evaluate | Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 21 | GitHub / GitHub Actions | Coding and deployment | Free public-repository automation | GitHub account already connected | Public actions free on standard runners; secrets must be protected | CI/CD, scheduled public metadata sync, source archive | Existing | External processing, retention, lawful access and provider outages; Supply-chain compromise, CI secrets, maintainer permissions and update integrity; Public repositories and CI logs can disclose source code or secrets; verify controls before handling sensitive data |
| 22 | GitHub Pages | Static hosting | Free public site | GitHub account already connected | Static only; cannot safely expose private API keys | Public research pages and approved demos | Existing | Public repositories and CI logs can disclose source code or secrets; verify controls before handling sensitive data |
| 23 | Graphify | Code graph | Open-source code intelligence | No for local version | Compute costs | Map code dependencies and retrieve related symbols | Existing | Licensing, data custody, permissions, dependencies and provider continuity; verify controls before handling sensitive data |
| 24 | n8n | Workflow automation | Source-available self-hosted edition | No SaaS signup if self-host | Hosting costs; licensing conditions apply | Scheduled retrieval, review queues and approvals | Evaluate | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; verify controls before handling sensitive data |
| 25 | LangGraph | Workflow engine | Open-source framework | No for code | Models and hosting separate | Stateful, auditable multi-step agent execution | Evaluate | Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; verify controls before handling sensitive data |
| 26 | OpenTelemetry | Monitoring | Open-source | No for self-host | Storage and service costs | Traces, measurements, performance logs | Evaluate | Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 27 | Google TimesFM | Forecasting | Open-weight research model | No for public code access | Hardware/model licence and data requirements | Time-series forecasts and baseline comparisons | Research | Incorrect forecasts, missing data, uncertain calibration and consequential over-reliance; verify controls before handling sensitive data |
| 28 | LocalAI | Local inference | Open-source self-hosted | No local account | Model download, compute and maintenance | OpenAI-compatible local endpoints, multiple model types | Compare endpoint security, VRAM and throughput | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 29 | vLLM | Local inference | Open-source self-hosted | No | Substantial GPU may be required; electricity/hosting | High-throughput inference server | Benchmark hardware requirements and concurrent requests | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 30 | SGLang | Local inference | Open-source self-hosted | No | GPU and deployment costs | Fast structured generation and model serving | Check compatible model and hardware | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 31 | MLX LM | Local inference | Open-source local | No | Requires compatible Apple Silicon hardware | Inference and adaptation on Macs | Compare local throughput and memory | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 32 | AirLLM | Local inference | Open-source local | No | Disk I/O can make large-model inference slow | Layer-streamed large-model inference | Measure tokens per second, disk usage and quality | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 33 | Colibrì | Local inference | Open-source candidate | No for source | Very low speed possible on laptop RAM | Experimental large-model CPU inference | Confirm published speed on real hardware and reproduce performance | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 34 | AnythingLLM | Local interface | Open-source/self-hosted | No local account | Model, embeddings and machine costs | Private document Q&A workspace | Check source provenance and local data retention | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 35 | LibreChat | Local interface | Open-source self-hosted | No provider login for self-host | Database, model gateway and server maintenance | Multi-provider chat UI, controls and workflows | Review permissions and integrations | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; verify controls before handling sensitive data |
| 36 | OpenRouter free models | Hosted inference | Free tier with request caps | Account and API key | 50 requests/day on free account; models/provider data policies vary | Compare multiple free models via one gateway | Record free model selection, latency and data terms | External processing, retention, lawful access and provider outages; Routing to alternate providers can change data jurisdiction and usage conditions; verify controls before handling sensitive data |
| 37 | Cerebras Inference | Hosted inference | Introductory free credit | Account and API key | Official pricing currently advertises $5 starter credit, not an unlimited tier | Short benchmark of high-speed inference | Credit exhaustion, model availability and privacy | External processing, retention, lawful access and provider outages; verify controls before handling sensitive data |
| 38 | NVIDIA NIM APIs | Hosted inference | Developer access or limited trial | NVIDIA account and credential | Provider quota and eligibility vary; production usage may cost | Model API comparison | Verify account-specific terms and limits | External processing, retention, lawful access and provider outages; verify controls before handling sensitive data |
| 39 | SearXNG | Search and ingestion | Open-source self-hosted | No local signup | Search engines may block or restrict automated queries | Metasearch discovery and source diversity | Respect provider access terms; measure recall | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; verify controls before handling sensitive data |
| 40 | Trafilatura | Search and ingestion | Open-source local | No | Web access subject to publisher restrictions | Clean article text extraction and metadata | Preserve original URLs and dates | Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; verify controls before handling sensitive data |
| 41 | Scrapy | Search and ingestion | Open-source local | No | Robots/access policies, compute and storage costs | Permissioned structured extraction | Test crawl politeness and rate limits | Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; verify controls before handling sensitive data |
| 42 | Apache Tika | Search and ingestion | Open-source local | No | Local compute, file-processing security risks | Extract text and metadata from public documents | Sandbox uploaded or untrusted files | Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; verify controls before handling sensitive data |
| 43 | Feedparser | Search and ingestion | Open-source local | No | Publisher feeds vary in availability and terms | RSS/Atom ingest of dated news items | Track original URL, author, update and correction | Publisher terms, manipulated pages, malicious content, unsupported citations and tracking; verify controls before handling sensitive data |
| 44 | Dify | Orchestration | Source-available self-hosted | No SaaS signup for self-host | Check licence/restrictions; model and hosting costs | Visual AI workflows and retrieval | Assess licence, permissions and data retention | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; verify controls before handling sensitive data |
| 45 | Flowise | Orchestration | Open-source self-hosted | No for local deployment | Model calls and hosting cost extra | Visual workflow and agent design | Benchmark logs, approvals and security | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; verify controls before handling sensitive data |
| 46 | Haystack | Orchestration | Open-source framework | No | Models and infrastructure separate | Composable evidence-first RAG pipelines | Measure citation fidelity and retrieval completeness | Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; verify controls before handling sensitive data |
| 47 | PydanticAI | Orchestration | Open-source framework | No for library | Underlying model APIs may cost | Type-checked structured outputs and agent tools | Test schema validation and failure modes | Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; verify controls before handling sensitive data |
| 48 | LlamaIndex | Orchestration | Open-source framework components | No for source | Cloud and model services may charge | Document indexing, retrieval and tool integrations | Check package licence and citations | Privilege escalation, prompt injection, unexpected tool calls and cross-provider data forwarding; verify controls before handling sensitive data |
| 49 | Langfuse | Observability | Open-source self-hosted | No for self-host | Storage, hosting and telemetry privacy costs | Trace model calls and prompt versions | Protect confidential prompts and personal data | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 50 | Arize Phoenix | Observability | Open-source self-hosted | No | Compute and evaluator-model costs | Trace/evaluate answers and evidence retrieval | Audit evaluation reproducibility | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 51 | MLflow | Observability | Open-source self-hosted | No | Database and storage hosting | Experiment and model version tracking | Document data and model provenance | External processing, retention, lawful access and provider outages; Endpoint security, model provenance, unreviewed packages and local data exposure; Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 52 | Promptfoo | Evaluation | Open-source local CLI | No for local testing | Target API usage and some security tests may cost | Regression evaluation and safe red-team tests | Test hallucinations and prompt injection using synthetic cases | Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 53 | DeepEval | Evaluation | Open-source local | No for core library | Evaluator API calls may cost | LLM output evaluation with test fixtures | Validate judging metrics with humans | Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 54 | Evidently AI | Evaluation | Open-source local | No local account | Monitoring storage and infrastructure separate | Data quality, drift and monitoring | Compare metrics to real-world errors | Prompt/trace leakage, model-judge error, sensitive logs and telemetry retention; verify controls before handling sensitive data |
| 55 | NVIDIA SkillSpector | Security | Open-source scanner | No for local tool | May not detect all malicious behaviour | Scan third-party AI skills before use | Manual code and permission review still required | Supply-chain compromise, CI secrets, maintainer permissions and update integrity; Static checks miss some malicious behaviour; independent code review remains required; verify controls before handling sensitive data |
| 56 | OWASP GenAI Security Project | Security | Open security guidance | No | Assessment and mitigations require work | Threat models and AI application checklists | Map controls to own agent privileges | Supply-chain compromise, CI secrets, maintainer permissions and update integrity; verify controls before handling sensitive data |
| 57 | Cloudflare Workers | Deployment | Free usage tier | Cloudflare account | Free request/CPU limits; no unattended unlimited inference | Server-side API gateway to protect secrets | Apply quotas, access rules and budget limits | External processing, retention, lawful access and provider outages; Supply-chain compromise, CI secrets, maintainer permissions and update integrity; verify controls before handling sensitive data |
| 58 | Hugging Face Spaces | Deployment | Free basic CPU hosting tier | Hugging Face account | GPU/higher memory are paid or quota-controlled | Public demo front-end or limited CPU prototype | Check uptime, data privacy and limits | External processing, retention, lawful access and provider outages; Supply-chain compromise, CI secrets, maintainer permissions and update integrity; verify controls before handling sensitive data |
| 59 | DuckDB | Data management | Open-source embedded analytics | No | Local disk, compute and backups | Analyse large sets of news events and indicators | Test reproducible data transforms | Misconfigured access, backups, cross-user leakage and retention; verify controls before handling sensitive data |
| 60 | PostgreSQL | Data management | Open-source database | No for local | Hosting, backups and admin time | Durable structured evidence and audit records | Access controls and schema validation | Misconfigured access, backups, cross-user leakage and retention; verify controls before handling sensitive data |
| 61 | FAISS | Data management | Open-source local | No | Local vector-index memory and CPU/GPU costs | Semantic similarity and duplicate detection | Test false matches and attribution | Misconfigured access, backups, cross-user leakage and retention; verify controls before handling sensitive data |
| 62 | Whisper.cpp | Speech and accessibility | Open-source local | No | Model licences, compute and recording permissions | Transcribe approved audio/video sources | Check language accuracy, privacy and consent | Recording consent, transcript confidentiality and voice model licences; verify controls before handling sensitive data |
| 63 | Piper | Speech and accessibility | Open-source local | No | GPL-3.0 implementation; voice model licences and local compute | Text-to-speech for accessibility | Active successor to archived rhasspy/piper; review GPL requirements and voices | Recording consent, transcript confidentiality and voice model licences; verify controls before handling sensitive data |
Every technology is listed with foreseeable technical and information-handling risks. This does not imply that a named provider or its staff have acted improperly. Nationality, employee background or geography alone does not establish intent; assess published data-handling policies, jurisdiction, access controls, incident response, audits and verifiable contractual terms. Open-source software may reduce dependence on a service provider but does not eliminate malicious packages, inaccurate models or local security problems.
Data handling default: keep sensitive or unpublished information local where feasible, restrict tokens to the minimum permissions, avoid embedding credentials in publicly deployed pages, document third-party processing and test a supplier's revocation and deletion process. Never send personal case files or confidential research to a free model API solely because it is free.
The Business Accelerator: Salesforce, Einstein Analytics & Open AI Integration compares 23 platform components and integration paths with free local tools, developer environments, production licence limits, security requirements and staged scaling strategies. Salesforce Free Suite is separate from Developer Edition; free trials do not grant unrestricted production analytics or API access.
Retain existing GitHub infrastructure; audit current pages and existing Graphify CI; prepare source schemas, synthetic evaluation cases and a consent-based test dataset. Experiment with open-source local tools on suitable hardware before creating subscriptions.
Evaluate one free model API, one extraction provider only if needed, and one cloud storage/search service only if the existing environment is insufficient. Each connection must have documented purpose, allowed data, quota, provider terms, key custody and a revocation procedure. Do not publish API keys on GitHub Pages.
Implement a server-side model gateway, local source cache, explicit citation record, independent-source deduplication, request budget caps, failure fallback, data retention limits and human review for any consequential action. Compare providers using identical prompts and datasets.
Deploy only once evidence coverage, cost per request, latency, factual accuracy, uncertainty calibration, accessibility and responsible-use controls have been measured. Public-safety information remains accessible independent of commercial subscriptions.
A free alternative can only be described as comparable after task-specific evaluation. For the AI Public Adviser, measure source coverage, claim attribution accuracy, factual error rate, contradiction detection, refusal to guess when evidence is inadequate, response latency, privacy safeguards and total monthly operating cost.
Recommended comparison: local model vs free hosted model vs a stronger benchmark model, using the same set of public questions, evidence passages and blinded scoring. Publish individual failure cases and date all results.
Research hub · AI demonstrations · Societal resilience programme · Productivity log