Osaurus
Osaurus runs private local AI agents on your Mac — free and MIT-licensed, with $20 per workspace shared agents for teams.
If you're on an Apple silicon Mac and want agents touching real folders without shipping context to a vendor, Osaurus is the one to beat — free, MIT-licensed, and the context layer sits on your disk. The Windows 11 beta arrived September 28, 2026, so the Mac-only objection is finally softening. Teams should look at Workspaces: $20 per workspace per month for 5 seats, unlimited shared agents, and a shared $20 cloud-credit pool undercuts seat-priced assistants — but that rate is explicitly
Verified 4d ago · liveness 81/100 · cite: rightaichoice.com/tools/osaurus
- Mac power users who want agents reading and editing real folders without sending data to a vendor
- Developers building local-first macOS automation that runs shell, Python, or Node inside a sandbox
- Privacy-sensitive work where cloud models are acceptable only after names and account numbers are redacted
- Small teams (up to 5 seats) where one expert builds an agent and everyone else asks it questions
- Anyone without an Apple silicon Mac — the app requires macOS 15.5+
- Cross-platform teams needing a supported Windows release today, since Windows 11 is still beta
- People who want zero setup and a plain chat window with no Helpers, skills, or workspaces to learn
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 Osaurus if you need a supported non-Mac release today or you want a centrally hosted team admin console — shared agents run on your own Macs over a network you host, and Windows 11 is still beta.
Each workspace includes $20 of monthly cloud credits, and unused monthly credits do not roll over — heavy months don't bank toward light ones.
A flat $20 per workspace per month (introductory rate, $0 for individuals) covers 5 seats, unlimited shared agents, and $20 in monthly cloud credits — cheaper per seat than Cursor Teams or ChatGPT Business, and it's per workspace rather than per seat, so you pay the same for 2 people or 5. Yearly billing gets two months free. It won't fit organizations needing SSO, audit, and procurement, which sit on the custom Enterprise tier.
In short
Osaurus — Osaurus runs private local AI agents on your Mac — free and MIT-licensed, with $20 per workspace shared agents for teams. Best for Mac power users who want agents reading and editing real folders without sending data to a vendor, Developers building local-first macOS automation that runs shell, Python, or Node inside a sandbox, Privacy-sensitive work where cloud models are acceptable only after names and account numbers are redacted. Free to start; paid plans from $20/mo.
What's new in Osaurus
Checked 4 days agoAcross the latest 5 updates: 3 launches and 2 changelog entries.
Osaurus 0.25.21 released
Adds MiniCPM-V 4.7 + DSpark support, Qwen-Image-2.1-Turbo image bundles, and raises file_write limits by removing the 30,000-character cap.
Osaurus 0.25.20 adds calculate tool and image handling fixes
Ships an exact-math calculate tool, hardened phone pairing channel, expanded phone image generation, and MCP access to version-manager toolchains.
Osaurus Is Now on Windows 11
The local-first agent harness arrives on Windows 11 in beta, with local models, private cloud access and agents.
Raptor 0.6: A Small Agent Model on a New Base
Raptor 0.6 replaces the looping 0.5 release, rebuilt on the Spark-X2.5-4B base as a 3.7 GB agent model tuned for the Osaurus harness.
Raptor 0.5: A Small Agent Model for the Macs Everyone Actually Owns
Open 8B-A1B model tuned for automation and business work; 6.3 GB on disk and built to run on 8-16 GB Macs.
What people actually say about Osaurus — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
39 mentions across 4 sources (Hacker News, YouTube, GitHub, Lemmy) · researched Aug 27, 2026.
Average across the 4 sources that answered — each source counts once, not each post.
- +Truly offline — no telemetry, no data leaving your Mac
- +Runs 30% faster than Ollama on Apple Silicon (GitHub/HN)
- +Single shared memory across local and cloud models
- +Agent loop with folder watchers, voice control, parallel jobs
- +Ollama-compatible — drop-in for your existing Ollama workflows
- −Apple Silicon only — no Windows, Linux, or Intel Mac
- −Long generations (1000+ words) can hang or spin forever
- −Memory leak issues when used as a server backend
- −Tool permissions granted but plugins still don't fire
- −No official support channel; relies on GitHub issues
- • You pay for your own compute — Mac hardware with enough RAM for larger models
- • Cloud models (ChatGPT, Claude, Gemini) require their own API keys and usage costs
- • Time investment to set up and debug agent workflows
Viability Score
How well maintained and how widely used is Osaurus? 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 →Key Features
- Native Apple-silicon MLX runtime written in Swift, with no Python dependency
- Raptor 0.6 agent model on a Spark-X2.5-4B base, 3.7 GB, bundled in the full DMG
- Run open models locally via Ollama, MLX, LM Studio, or Apple Foundation Models
- Connect OpenAI, Anthropic, Google Gemini, OpenRouter, Venice, or Grok with your own API key
- Privacy Filter redacts names, emails, and account numbers from cloud prompts before sending
- International privacy filter presets with a region picker (v0.25.17)
- Autonomous Helpers read, search, and edit files inside a folder you point them at
- Shell, Python, and Node execution in an isolated sandbox
- Agents browse sites, read pages, and fill forms with a persistent browser
- Built-in web search with no API key required
- Agents run on a schedule or when a folder changes, with a morning daily summary
- Memory distills past conversations into facts surfaced only when relevant
- Self-loading skills with SKILL.md import from GitHub
- Per-agent cryptographic identity with scoped, revocable access keys
- MCP in and out, plus a native plugin registry with a stable ABI
About Osaurus
Osaurus is a native macOS app for Apple silicon that runs AI agents on your own machine instead of a vendor's cloud. Point a Helper at a folder — contracts, intake notes, bank exports — and it reads, searches, and edits the files inside, asking for approval before each change and staying inside the folder you chose, with undo available. Shell, Python, and Node run in an isolated sandbox. The model is a setting, not a lock-in. Connect OpenAI, Anthropic, Google Gemini, Venice, OpenRouter, or Grok with your own API key, or run open models locally through Ollama, MLX, LM Studio, or Apple Foundation Models. Osaurus ships its own MLX runtime in native Swift (no Python) and its own small agent model, Raptor 0.6, on a Spark-X2.5-4B base — a 3.7 GB download bundled in the full DMG, so chat and file work run offline right after install on 8 GB Macs and up. When a prompt does go to a frontier model, the Privacy Filter spots names and account numbers, shows you the redactions, and lets you remove them before anything is sent; Insights logs everything that left your Mac. The layer around the model is what you actually keep: memory distilled from past conversations into facts surfaced only when relevant, self-loading skills with SKILL.md imports from GitHub, per-agent cryptographic identity with scoped revocable keys, MCP in and out, a plugin registry, and a local server speaking OpenAI, Anthropic, Open Responses, and Ollama APIs on one port so existing SDKs keep working. Agents browse with a persistent browser, run on schedules or folder-change triggers, and hand you a morning summary. Individuals get the full app free, forever, with no account, installable via brew install osaurus on macOS 15.5+. Teams pay $20 per workspace per month (introductory rate; monthly or yearly with two months free) for up to 5 seats, unlimited shared agents, and a $20 monthly cloud-credit pool. A Windows 11 beta shipped September 28, 2026. Against cloud-first assistants like Claude or ChatGPT,
Behind the Verdict
Pick Osaurus when the work is local by nature: a folder of contracts, intake notes, or a month-end bank export that an agent can read, reconcile, and redline without the contents leaving the laptop. The approval gate before file edits and the tamper-evident Insights log are the two features we'd actually check first — they're what makes a folder-scoped agent tolerable in a practice or finance context. We'd also reach for it as a harness rather than a chatbot. The local server speaks OpenAI, Anthropic, Open Responses, and Ollama APIs on one port, so tools you already wired up keep working, and MCP runs in both directions. If you flip between local and cloud models depending on the sensitivity of the task, the Privacy Filter plus the Cloud browser and Credits menu in the redesigned picker (v0.25.16) make that a two-click decision instead of a config file. Where it bites: there is real setup. Helpers, skills, memory, and workspaces are concepts you have to learn, and the app assumes macOS 15.5+ on Apple silicon. Cross-platform teams get a Windows 11 beta as of September 28, 2026 — usable, not a supported release. And there's no centrally hosted admin console; shared agents run on your team's Macs over a network you host, which is the point but also a real operational chore. The closest alternative for most buyers is just paying for a cloud assistant and accepting the data flow. Osaurus is the argument against that: free forever for one person, MIT-licensed, no account, and cloud models optional rather than mandatory. If your team's knowledge already lives in one expert's head, the $20 workspace with unlimited shared agents is where the math flips.
Researching Osaurus? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Osaurus actually fits — and what changes day-one when you adopt it.
Installs with brew install osaurus, points an agent at a repo folder, and lets it read, search, and edit files while running Python and Node in the sandbox to test changes.
Outcome: Coding tasks run with local models or a frontier model through the Privacy Filter, and every agent keeps its memory and skills when the model is switched.
Loads contracts into a pointed folder, uses the Privacy Filter with an international region preset to scrub names and emails from prompts, and reviews the redactions before approving the send.
Outcome: Cloud models stay usable for drafting while personal details never leave the Mac, and the tamper-evident Insights log records what happened.
Builds a Release Assistant agent once, shares it in a $20/workspace Workspaces setup, and sets schedules so a daily summary is ready each morning.
Outcome: Sales and support ask the agent instead of the lead, shared agents run over the encrypted network the team hosts, and the $20 monthly credit pool covers cloud calls by default.
Use Cases
- Create an autonomous coding agent that reads, writes, and tests scripts on your Mac.
- Run a personal knowledge assistant whose memory persists across model switches.
- Automate file organization and folder monitoring with natural-language commands.
- Build a voice-controlled agent that executes local tools and scripts.
- Turn plain English into macOS automation with AppleScript models.
- Run AI entirely offline for sensitive data or travel, using the bundled Raptor model.
- Import existing conversations from ChatGPT, Claude, Grok, or Gemini and continue locally.
- Use the Orchestrator to delegate a task between specialist agents.
Models Under the Hood
as of 2026-10-17
Limitations
- Osaurus requires macOS 15.5+ on Apple silicon; the Windows 11 build is still in beta, and no Intel Mac, Linux, or older-Windows support is mentioned.
- Local model quality and speed depend on your Mac's hardware.
- Cloud options (Anthropic, OpenAI, Google, Venice AI and others) require your own API keys or prepaid credits and send selected prompts off-device, with the Privacy Filter redacting personal details first.
- Workspaces team plan is $20 per workspace per month (introductory rate) for up to 5 seats and unlimited shared agents, while local inference stays unmetered.
as of 2026-10-04
Verification history
We have re-verified Osaurus 9 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.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 9 verification passes.
Free to cite with attribution — this page re-verifies continuously.
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.
Plans compared
For each published Osaurus tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Free
$0
Ideal for
Solo Mac user on Apple silicon who wants agents over local files with no account and no subscription
What this tier adds
Starting tier: full MIT-licensed app, local models with no usage charges, bring-your-own cloud keys, MCP, skills, memory, and scheduled agents
Workspaces
$20 per workspace/mo (introductory rate)
Ideal for
Team of up to 5 who want one expert's agents shared with everyone over a network they host
What this tier adds
Adds unlimited shared agents, roles, invite links, and a billing portal, plus $20 of cloud credits per workspace per month at an introductory $20 per workspace
Enterprise
Custom
Ideal for
Organizations with procurement, security review, and scale requirements that outgrow the 5-seat workspace model
What this tier adds
Adds volume pricing and annual invoicing, security review and procurement support, SSO, admin controls and audit, and a named support contact
Where the pricing makes sense
The company stage and team size where Osaurus's pricing actually pencils out — and where peers do it cheaper.
A flat $20 per workspace per month (introductory rate, $0 for individuals) covers 5 seats, unlimited shared agents, and $20 in monthly cloud credits — cheaper per seat than Cursor Teams or ChatGPT Business, and it's per workspace rather than per seat, so you pay the same for 2 people or 5. Yearly billing gets two months free. It won't fit organizations needing SSO, audit, and procurement, which sit on the custom Enterprise tier.
Setup time & first value
How long it actually takes to get something useful out of Osaurus — broken out by persona, not the marketing-page minute.
Individuals: a brew install osaurus on macOS 15.5+ gets you running in minutes, and the 3.6 GB full DMG bundles Raptor 0.6 so chat works offline right after install with no model download. Teams: a Workspaces trial converts in 14 days with a card at checkout; roles, invite links, and billing all live in the app, and shared agents run over a network you host with no server or config file to stand
Switching to or from Osaurus
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From ChatGPT, Claude, Grok, or Gemini: import existing conversations into Osaurus and continue them locally with a model of your choice.
- →From a hosted AI assistant subscription: point an agent at your local folders and run the same prompts locally or through the Privacy Filter on the cloud path.
- →From a coding-only tool: reuse the same repo folder with shell, Python, and Node sandboxed, then share the agent via Workspaces for up to 5 seats.
- ↗To a centrally hosted team assistant: Workspaces agents run on your own Macs, so an admin-console-based product like ChatGPT Business is the switch if you need that.
- ↗To an editor-first workflow: if you only want inline code completion and chat in an IDE rather than folder-wide agents, Cursor covers that ground.
Integrations
Resources & Guides
- Guideosaurus.ai
Guides · Osaurus
In-depth how-to from osaurus.ai
- Resourceosaurus.ai
Blog · Osaurus
Helpful link from osaurus.ai
- Resourceosaurus.ai
Models · Osaurus
Helpful link from osaurus.ai
- Resourceosaurus.ai
Skills · Osaurus
Helpful link from osaurus.ai
- Resourcegithub.com
Osaurus · Osaurus
Helpful link from github.com
Tutorials & Learning
YouTube returned 6 videos for “Osaurus”, and we withheld 6: 6 could not be judged, because “Osaurus” 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 Osaurus.
Official links
Tools that pair well with Osaurus
Common stack mates teams adopt alongside Osaurus, with the specific reason each pairing earns its keep.
Localforge
Localforge is a free, MIT-licensed desktop coding agent GUI that runs autonomous multi-file code edits in your own repo with any LLM.
OpenAI Agents SDK
OpenAI Agents SDK: a free, MIT-licensed Python framework for building multi-agent workflows with handoffs, guardrails, and tracing.
Cherry Studio
Cherry Studio runs 300+ cloud and local AI models in one free, open-source desktop client
Featured Head-to-Head Comparisons
Osaurus vs Locus Robotics
Locus Robotics and Osaurus serve entirely different markets. Locus is a warehouse robotics platform for high-volume physical fulfillment, while Osaurus is a local AI agent toolkit for macOS developers. Choose based on your domain: if you run a warehouse needing flexible automation, Locus is the proven choice; if you're a developer wanting private, autonomous AI on your Mac, Osaurus is free and powerful.
Osaurus vs Presto Voice
Buy Presto Voice if you operate a QSR chain and need to boost drive-thru revenue via AI—its upselling engine and high non-intervention rate are proven. Choose Osaurus if you're a macOS developer wanting to run local AI agents with full privacy and control. These tools serve entirely different needs; the choice is dictated by your environment (restaurant vs. desktop) and your budget (enterprise vs. free).
Osaurus vs Truleo
Truleo and Osaurus are incomparable tools serving entirely different domains. Truleo is a paid, specialized intelligence platform for law enforcement agencies needing to connect siloed data sources and automate lead generation. Osaurus is a free, open-source local AI agent framework for macOS developers who prioritize privacy and offline capability. Your choice depends entirely on whether you are a police department or a privacy-conscious Mac power user.
Alternatives to Osaurus
View all
Localforge
Localforge is a free, MIT-licensed desktop coding agent GUI that runs autonomous multi-file code edits in your own repo with any LLM.
OpenAI Agents SDK
OpenAI Agents SDK: a free, MIT-licensed Python framework for building multi-agent workflows with handoffs, guardrails, and tracing.
Cherry Studio
Cherry Studio runs 300+ cloud and local AI models in one free, open-source desktop client
Frequently Asked Questions
Best-of guides
Used Osaurus? Help shape our editorial sentiment research.