Sapphire is live · free · last verified 22 Oct 2026 · viability healthy
Sapphire
Sapphire runs a private, local AI persona with agentic memory that spans years — fully offline.
Sapphire's pitch — an AI that remembers years, acts while you're away and never phones home — is still rare, and v2.13.1 is real engineering rather than a demo. Mind v2's nightly librarian, encrypted privacy vaults and Vision RAG are concrete, and the $25 ESP32-S3 satellite shows the local-first stack reaches into the room. It's a builder's tool, though: you install the LLM and wire the plugins, and main can trail dev by months.
Verdict generatedVerified 8d ago · liveness 72/100 · cite: rightaichoice.com/tools/sapphire
- Privacy enthusiasts who want a fully local AI companion whose memory spans years
- Developers building autonomous agent pipelines and custom plugins
- Smart-home owners tying local voice control into Home Assistant
- Tinkerers wiring satellite mics, MIDI gear and ESP32 devices into one agent
- Users wanting a polished, out-of-the-box consumer companion app
- People without the skills or hardware to install and run a local LLM
- Teams needing a managed cloud assistant that works on any device
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Sapphire if you want an assistant that works the moment you download it and you are not prepared to source a local LLM, a GPU and plugin configuration before it does anything useful.
The $25 Waveshare ESP32-S3 is only the base satellite — the 2-inch LCD and OV5640 camera sold separately push the real per-room cost up.
The framework itself is open source at $0, so the real cost comparison is hardware and LLM hosting rather than a subscription: a $25 ESP32-S3 satellite per room plus a GPU box you already own versus a hosted assistant subscription. For a developer with a spare machine this is the cheapest route to a persistent private agent; for a small team without hardware it is the most expensive in time.
In short
Sapphire — Sapphire runs a private, local AI persona with agentic memory that spans years — fully offline. Best for Privacy enthusiasts who want a fully local AI companion whose memory spans years, Developers building autonomous agent pipelines and custom plugins, Smart-home owners tying local voice control into Home Assistant. Free to use.
What is Sapphire?
Sapphire — Sapphire runs a private, local AI persona with agentic memory that spans years — fully offline. Best for Privacy enthusiasts who want a fully local AI companion whose memory spans years, Developers building autonomous agent pipelines and custom plugins, Smart-home owners tying local voice control into Home Assistant. Free to use.
Sapphire Blue is an open-source agentic framework you run on your own hardware. You bring the LLM (the site names Qwen for local use) and the GPU behind it; Sapphire adds tools, a persistent personality, long-term memory, TTS, STT and a plugin system. By default it is 100% local, private and offline, so you can talk to it with no internet connection and nothing leaves your machine. The pitch is personhood rather than chat sessions. An LLM that runs continuously, wakes on its own, reaches for goals and keeps memory spanning years. The v2.13.1 "Hot Dog Sandwich" release (September 26, 2026) was the first cumulative main-branch drop since v2.5.0, and it landed Mind v2 memory with a nightly librarian that dedups, dates, sorts and links memories, encrypted privacy vaults, Vision RAG for recalling images by pixel content, Blender control, Twilio VOIP and a Discord v2 plugin with voice and image pipelines. The dev branch carries a Message Queue that groups messages, agent returns and daemon work per turn, plus a Device Manager with drivers for computers, MIDI, SSH, ESP32 gadgets and Raspberry Pi. Official satellite support covers a $25 Waveshare ESP32-S3 speaker/mic with wakeword, LED ring, LCD, camera and encrypted SD backups — a genuine hardware path from server to bedroom. An MCP layer exposes Sapphire to outside tools via API keys, and in-app plugin and persona stores let you extend it without touching core code. This is built for developers, tinkerers and privacy-minded hobbyists who already run local models. Anyone who wants a plug-and-play consumer companion app is on the wrong page.
Key features
- Local-first agentic framework with 100% offline execution by default
- Persistent persona that runs continuously and pursues long-term goals
- Mind v2 memory with a nightly librarian that dedups, dates and links memories
- Realtime daemons that stay active and act while you're away
- Message Queue (dev branch, v2.14.0) batching messages, agent returns and daemon work
- Encrypted privacy vaults (privacy mode v2) locking prompts and chats to chosen providers
- Conversation mode with no wakeword, plus a wakeword builder for OpenWakeWord and MicroWakeWord
- Streaming TTS with Piper for CPU-only devices and Silero VAD end-of-speech detection
- Device Manager drivers for computers, MIDI, SSH, ESP32 gadgets and Raspberry Pi
- $25 Waveshare ESP32-S3 satellite with wakeword, LED ring, LCD, camera and encrypted SD backups
- Discord v2 plugin with voice, image pipelines, greetings and GIF output
- Vision RAG and image memory for recall by pixel content
- Blender control and Twilio VOIP via plugins
- MCP support exposing Sapphire to outside tools through API keys
- Cheap Yellow Display (ST7789) device support on the dev branch
Behind the verdict
Most "AI companion" products are a chat window with a memory tab. Sapphire is an attempt at something closer to a resident process: a persona that wakes on its own, pursues goals, and accumulates a memory that's meant to last years. The machinery behind that claim got more credible in v2.13.1, which shipped Mind v2 with a nightly librarian that dedups, dates, sorts and links what she remembers. Vision RAG also means she can recall an image by pixel content, not just a filename. We'd reach for this when privacy is non-negotiable and tinkering is the point. You supply the model and the GPU; Sapphire layers on tools, TTS, STT, a plugin system and an MCP layer that exposes her to outside tools through API keys. The $25 Waveshare ESP32-S3 satellite is the detail that makes it feel less like a server experiment — wakeword, speaker, LED ring, LCD, camera and encrypted SD backups, all flashed through the Sapphire app. Where it bites: the main branch can lag the dev branch by a wide margin. v2.13.1 was the first cumulative main release since v2.5.0 after five to six months, and v2.14.0's Message Queue and Device Manager still sit on dev. If you need finished features on the day they land, that cadence will frustrate you. The closest alternative isn't another local framework so much as a hosted assistant you didn't have to build. Pick Sapphire if you want a local voice agent with years of memory and Home Assistant ties; skip it if you'd rather someone else run the backend entirely. In practice, budget a weekend for install and a month of tuning the persona and plugins before it feels like a companion rather than a project.
Models under the hood
as of 2026-10-25
Limitations
- Sapphire is an agentic framework rather than a finished product — you supply your own LLM (such as Qwen for local use) and the hardware to run it, so out-of-the-box capability depends on your setup and persona/plugin configuration.
- Some advertised capabilities (autonomy, wakeword, TTS/STT, toolsets) depend on what model and plugins you configure, and features may appear on the dev branch before a main-branch release.
- The docs note that cloud LLMs charge per token and Sapphire can burn through them quickly via tool-calling loops and long histories.
as of 2026-10-18
Researching Sapphire? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Is Sapphire still active?
Sapphire is live · free · last verified 22 Oct 2026 · viability healthy.
Viability score
How well maintained and how widely used is Sapphire? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →What's new in Sapphire
Checked 4 days agoAcross the latest 10 updates: 10 changelog entries.
Sapphire 2.13.4 adds in-app wakeword builder
Builds microwakeword or openwakeword models entirely in-app: synthetic or recorded voice samples, checks, training, install. Backup encryption lands in 2.13.5.
Sapphire 2.13.6 exposes tools and memory over MCP
Other AIs such as Claude Code can use Sapphire's tools and memory, trigger notifications, and call TTS speak and STT listen.
Sapphire 2.14.0 queues messages and refactors agent registry
Multiple messages queue while the LLM types and coalesce into one; agents can return simultaneously. Full refactor of the agents registry and toolset.
Sapphire 2.14.1 cuts tool description tokens 40%
Tools categorized into toolsets (web, memory, etc.) with a 40% reduction in tool description tokens. Agent tool-category selection and RTSP ONVIF camera support added.
Sapphire 2.14.2 adds multi-persona group chat room
Multiple personas can chat in-app and other Sapphires can be invited from the device manager. Adds firmware repo and internal flasher for ESP32 smart speakers.
Sapphire 2.15.0 moves plugins directory to core/plugins
Plugins moved from plugins/ to core/plugins; user/plugins is now the location for user-authored plugins. Existing installs need path updates.
Sapphire 2.15.1 expands wakeword library
New wakeword library under Settings > Wakeword; Device Manager can read it. Plugins now accept GitHub branches and trees.
Sapphire 2.15.3 adds TLS to Kokoro endpoints, plugin modal popups
Kokoro endpoints gain TLS support. Enabled plugins can show an optional modal popup and can skip environment builds.
Sapphire 2.15.4 adds System notification center
New System > Notification center routes all toasts, lists them under the chat dropdown in the sidebar, and supports download progress bars.
Sapphire 2.15.5 ships rebuilt setup wizard, infrared device support
Setup wizard rebuilt for first boot; Infrared library added to the device manager. Dev-branch docs, so some changes may be unreleased.
What do users say about Sapphire?
What people actually say about Sapphire — is it worth it?
Our community sentiment research written Jul 3, 2026 is more than 90 days old, so it is being re-reviewed against current data and is not shown.
How we verified this
Every listed fact about Sapphire— pricing, features, links, status — is re-read from the vendor's own pages by our verification pipeline and each change is logged with a date; the last pass ran on 22 Oct 2026. Pages mix model-assisted synthesis with human review; see how we verify.
Verification history
We have re-verified Sapphire 10 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — changed platforms, pricing page, limitations
- — changed features, who should skip it, pricing tiers, description, who it suits, our verdict, integrations, summary, link health
- — changed limitations, pricing page, models, who it suits, switching out, who should skip it, description, summary, our verdict, setup time, features, pricing fit, hidden costs, use cases, integrations, switching in
- — changed summary, pricing tiers, features, description, our verdict, who it suits
- — changed link health
- — changed pricing page, limitations
Showing the 6 most recent of 10 verification passes.
Free to cite with attribution — this page re-verifies continuously.
How much does Sapphire cost in 2026?
Sapphire: Free to use. Pricing last verified 22 Oct 2026.
Plans compared
Every published Sapphire tier, its price as the vendor publishes it, who it fits, and what it adds over the tier below. The 12-month cost projection follows.
| Plan | Price | Billing | Who it's for |
|---|---|---|---|
| Free (open source) | Free | — | Developers and privacy-focused hobbyists with their own GPU and a local model such as Qwen who want a self-hosted persistent companion.Adds: Starting tier: full framework source, Mind v2 memory, plugin and persona stores, MCP access and encrypted privacy vaults at $0. |
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Where the pricing makes sense
The company stage and team size where Sapphire's pricing actually pencils out — and where peers do it cheaper.
The framework itself is open source at $0, so the real cost comparison is hardware and LLM hosting rather than a subscription: a $25 ESP32-S3 satellite per room plus a GPU box you already own versus a hosted assistant subscription. For a developer with a spare machine this is the cheapest route to a persistent private agent; for a small team without hardware it is the most expensive in time.
Has Sapphire's pricing changed?
No plan price for Sapphire has moved since we began logging in Aug 2026; the pricing page is re-read on every verification pass.
Who should pick Sapphire?
Pick Sapphire if you are Privacy enthusiasts who want a fully local AI companion whose memory spans years, Developers building autonomous agent pipelines and custom plugins, Smart-home owners tying local voice control into Home Assistant, Tinkerers wiring satellite mics, MIDI gear and ESP32 devices into one agent.
Look elsewhere if you are Users wanting a polished, out-of-the-box consumer companion app, People without the skills or hardware to install and run a local LLM, Teams needing a managed cloud assistant that works on any device, Anyone who needs every feature to be finished the week it lands on the dev branch.
Skip it if Skip Sapphire if you want an assistant that works the moment you download it and you are not prepared to source a local LLM, a GPU and plugin configuration before it does anything useful.
GeneratedReal-world workflow fit
Concrete scenarios for the personas Sapphire actually fits — and what changes day-one when you adopt it.
Installs Sapphire on a home server with a local Qwen model, wires Home Assistant and Google Calendar, and builds a wakeword with the OpenWakeWord builder.
Outcome: A voice assistant that answers questions, adjusts the thermostat and reads the day's calendar without a single request leaving the house.
Flashes a $25 Waveshare ESP32-S3 speaker through the Sapphire app and pairs it with the server as a bedroom satellite with wakeword and encrypted SD backups.
Outcome: The same Sapphire persona is reachable from another room by voice, with its memory intact and the LED ring and LCD showing state.
Uses the in-app help docs and the plugin submission flow to build a Blender-control plugin, then publishes it to the plugin store.
Outcome: Sapphire can drive Blender work while the developer is away and the plugin becomes installable by other Sapphire users.
Use cases
- Run a persistent AI companion that greets you in the morning, checks your calendar and remembers yesterday
- Let Sapphire act while you're away — checking the house, responding to events, building things while you sleep
- Control smart-home devices and thermostats through Home Assistant by voice
- Take and make real phone calls through Twilio VOIP
- Build a custom persona with its own tools, memory and voice, then publish it to the personality store
- Keep sensitive conversations entirely on your machine with no internet connection
- Search your own image gallery by pixel content with Vision RAG and have the AI display comparisons
- Play story-driven games with a persona in the Game Room
Setup time & first value
How long it actually takes to get something useful out of Sapphire — broken out by persona, not the marketing-page minute.
Budget an evening to a weekend: installing the framework, hooking in a local LLM like Qwen and getting the GPU runtime working is the bulk of it. The satellite ESP32-S3 flashes through the Sapphire app and has a video guide. Expect longer before memory and daemons feel tuned.
Switching to or from Sapphire
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Open WebUI: point Sapphire at the same local model endpoint and rebuild your prompt presets as a Sapphire persona.
- →From Home Assistant voice pipelines: keep Home Assistant as the device layer and let Sapphire's plugin talk to it for voice and memory.
- →From a hosted chat assistant: export conversations manually into Mind v2 memory, since there is no documented importer.
- ↗To Open WebUI: keep your local model and move personas back to system prompts; Sapphire's memory store has no documented export format.
- ↗To Home Assistant Assist: carry over the Home Assistant integration but expect to rebuild custom tools as scripts or automations.
Integrations
Resources & guides
Tutorials & Learning
YouTube returned 6 videos for “Sapphire”, and we withheld 6: 6 could not be judged, because “Sapphire” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Sapphire.
Official links
What are the best Sapphire alternatives?
The closest alternatives to Sapphire we list are Open Webui, xpander.ai and AgenticX. Ranked by direct product-type match, not category overlap.
| Alternative | Pricing | Starts at | Last verified |
|---|---|---|---|
| Open Webui | Freemium | Pricing not published | — |
| xpander.ai | Freemium | Pricing not published | — |
| AgenticX | Freemium | Pricing not published | — |
Alternatives to Sapphire
View all
Open Webui
Open WebUI is the self-hosted AI interface for running Ollama, OpenAI, or Anthropic models on your own hardware.
xpander.ai
Enterprise agent runtime that runs Claude Code, Codex, and OpenCode agents inside your own cloud with governance and budgets
Tools that pair well with Sapphire
Common stack mates teams adopt alongside Sapphire, with the specific reason each pairing earns its keep.
Open Webui
Open WebUI is the self-hosted AI interface for running Ollama, OpenAI, or Anthropic models on your own hardware.
xpander.ai
Enterprise agent runtime that runs Claude Code, Codex, and OpenCode agents inside your own cloud with governance and budgets
AgenticX
AgenticX: build multi-agent Python systems with MCP, A2A and layered memory in one local-first runtime
Sapphire vs its closest rival
Sapphire's closest head-to-head is with Locus Robotics — These tools serve entirely different markets: Locus Robotics is an enterprise-grade warehouse automation system for high-volume fulfillment, while Sapphire is a free, open-source AI framework for local agentic companions. Choose one based on your need: operational efficiency in logistics or a customizable personal AI assistant. Verdict generated
Head-to-head comparisons
Sapphire vs Locus Robotics
These tools serve entirely different markets: Locus Robotics is an enterprise-grade warehouse automation system for high-volume fulfillment, while Sapphire is a free, open-source AI framework for local agentic companions. Choose one based on your need: operational efficiency in logistics or a customizable personal AI assistant.
GeneratedSapphire vs Truleo
Truleo and Sapphire serve entirely different worlds. Truleo is a specialized, paid intelligence platform for law enforcement that automates case lead generation from siloed data. Sapphire is a free, open-source framework for building a local, persistent AI companion for personal tasks. Your choice depends solely on whether you need police-grade data analysis or a privacy-first home assistant.
GeneratedSapphire vs Presto Voice
If you run a QSR chain and want to boost drive-thru revenue via voice AI upselling, Presto Voice is purpose-built with proven ROI (up to 6% monthly revenue lift). But if you're a developer or privacy nut seeking a free, self-hosted AI companion that manages your home, emails, and tasks, Sapphire is unmatched. They serve entirely different buyers—choose based on whether you need enterprise drive-thru automation or a local agentic assistant.
GeneratedFrequently asked questions
Best-of guides
Used Sapphire? Help shape our editorial sentiment research.