inthosAIWant Mythos? Try Winthos!
Specialist frontier-class AI starting at $5/month for global accessibility where $20/month subscriptions lock people out.
At Winthos, we aim to win (together)!
Operated commercially by Win Empire LLC (Florida Sunbiz L26000313506, DBA Winthos AI) across United States (US) and European Union (EU) data centers, paired with public-charity security and humanitarian tools from WIN'S VPN INC. (Florida Sunbiz N26000008302, IRS 501(c)(3) / 509(a)(2)). User-selectable US and EU datacenter routing is coming soon so individuals and enterprise teams choose where prompts and dedicated nodes run.
AI built to boost human productivity, not replace it. Scan and remediate code with Winthos Cyber first at low cost before expensive generalist passes.
MEET THE WINTHOS FAMILY
Specialist models operated across United States (US) and European Union (EU) data centers, with speed and context badges bound to our centralized model configuration (135.2 tok/s stream · 5,000 tok/s cluster, 185 ms TTFT).
Offline local health, posture coaching, agriculture, and remote survival assistant. Designed for on-device CPU/GPU/NPU auto-tuned execution inside Winthos AI Chat & Anti-Virus alongside Real-Time Deepfake Detection.
Explore the security & health suiteCompact 1M-context cybersecurity model built to inspect codebases, threat logs, and synthetic deepfake signal patterns. Planned open-weight release on Hugging Face at 10M+ community followers.
Open-weight release is conditional on that community milestone. Public weights are not offered on this page.
Explore the Cyber familyClinical, medical, and U.S. legal analysis model supporting clinicians, researchers, and attorneys across complex records and timelines. Casual web preview chats use a 64K to 128K token budget; up to 1M context is planned for Dedicated Cloud GPU, API, and Max tiers.
Clinical and legal specialist model; licensed physician CoT validation planned. Regulatory authorization is not established here.
Explore Winthos SuperFlagship model for software engineering, authorized security testing, and vulnerability remediation. Scan and patch code vulnerabilities with Winthos Cyber first at low cost before running expensive final generalist passes.
Benchmark statusModel specifications and Winthos Health Mini availability are supplied by Winthos. Winthos benchmark scores are withheld; source-reported reference results are listed separately below.
THE WINTHOS ECOSYSTEM
Commercial software innovation from Win Empire LLC (DBA Winthos AI) paired with humanitarian public-charity infrastructure from WIN'S VPN INC. (IRS 501(c)(3) / 509(a)(2)), uniting nine products under one mission.
Winthos Super, Winthos Cyber, Winthos Cyber Mini, Winthos Health Mini) operated commercially by Win Empire LLC (DBA Winthos AI) across US and EU data centers.Winthos Cyber first at low cost before running expensive final generalist passes.Winthos Health Mini auto-detects device chips across GPU or NPU, supports remote offline communities (first aid, agriculture, survival), local MRI/history/wearable review, live webcam posture coaching, specialist care coordination, and enterprise clinician triage.<think>...</think> reasoning formats.WIN'S VPN INC., the public charity behind Win's VPN, holds active IRS 501(c)(3) / 509(a)(2) status, while Win Empire LLC (DBA Winthos AI) operates commercial AI and software services designed to reinforce global access. We target January 2027 EU expansion via a Southern France accelerator alongside US and EU GPU partnerships.
PUBLIC SAMPLE TEASER DECK
Explore the 8-slide public overview using the step tabs, Previous/Next buttons, or keyboard Left and Right arrow keys. Proprietary hardware, protocol internals, and unit economics remain gated in the confidential partner and investor deck.
SLIDE 01 / VISION
Pairing commercial AI innovation (Win Empire LLC, DBA Winthos AI, Florida Sunbiz L26000313506) with public-charity humanitarian infrastructure (WIN'S VPN INC., Florida Sunbiz N26000008302, IRS 501(c)(3) / 509(a)(2), approved in 8 days).
SLIDE 02 / THE PROBLEM
High-cost AI ($20 to $200/month), $6,000+ legal fees, complex small-business licensing, and digital threats lock out most of the world, while 50M people in Myanmar and millions globally face censorship, spyware, and limited medical access.
SLIDE 03 / SPECIALIST AI STARTING AT $5/MONTH
Purpose-built models (Winthos Cyber, Winthos Super, Winthos Health Mini) engineered for software security, clinical/legal analysis, and local offline execution at a fraction of generalist frontier inference cost.
SLIDE 04 / END-TO-END SECURITY STACK
Patent-pending post-quantum Winn Protocol v3 VPN with calculator stealth disguise, Rust offline Winthos AI AV Engine with real-time deepfake detection, automated vulnerability patching, portable connected-device threat hunting, and Phish Proof Mail OIDC SSO.
SLIDE 05 / EDUCATION, SMALL BUSINESS & LEGAL ACCESS
Free Winthos University campus (5 accelerated BS+MS tracks with VR labs and oral exams), $50 StartBusiness.tech licensing and direct-ordering sites, and $500 AI permanent residency preparation.
SLIDE 06 / CULTURAL & CONSUMER FLYWHEEL
Paw Kart Tour (live on Microsoft Store and Mac) teaches automotive engineering and UNESCO heritage appreciation to families worldwide, driving organic discovery across the Winthos family.
SLIDE 07 / SHIPPED TRACTION
Live multi-OS VPN apps and Chrome extension, live Winthos AI US/EU preview, live Winthos University campus shell, live Paw Kart Tour release, USPTO provisional and non-provisional Winn Protocol filings, and 20 owned domains.
SLIDE 08 / PARTNER, ACCELERATOR & INVESTOR ACCESS
Targeting January 2027 EU expansion via Southern France accelerator plus US and EU GPU partnerships. Qualified accelerators, compute partners, and institutional investors can request the full confidential deck below.
LIVING CULTURAL HERITAGE
Human wisdom earned across millennia meets accessible modern tools. Explore our animated Bagan sunrise and Parthenon evening tributes here or read the full public-charity mission and policy essays on Mission & Charity.
Evolution = (AI: All Knowledge, Infant Thinking) + (Human: Life-Earned Critical Chain-of-Thought).
Read our missionWHAT WILL YOU MAKE POSSIBLE?
Your experience brings the context. AI helps you explore the possibilities. You stay in charge.
Understand an unfamiliar topic. Find the words you were looking for. Break a big project into a first step you can actually take.
Explore the security suiteA community garden, from first thought to first season.
Who will the garden serve, and what would a useful first season look like?
Map the space, water access, budget, and people willing to help.
Draft a one-page proposal for neighbors to discuss.
WINTHOS AI CHAT & ANTI-VIRUS
A unified security suite with AI conversation, an integrated AI IDE, a post-quantum self-deleting messenger, Winthos Shredder, and Real-Time Deepfake Detection (Audio, Video & Image Deepfake Verification) at its heart.
Winthos Health Mini is designed for local CPU/GPU/NPU execution in the app, while the Rust security engine flags synthetic voice clones, manipulated video calls, and AI-generated phishing media in real time. Supported builds, context limits, offline operation, and voice output require verification in the app's release documentation.
The app is designed to load the model when needed and eject it to release memory for other work. Device, model build, and context settings can affect memory use; measurements and release-build verification are pending.
Winthos Health Mini download informationThe free Winthos AI Antivirus, Firewall & Deepfake Detection Engine is in development and scans files on Android today alongside real-time audio, video, and image deepfake verification and Winthos Health Mini chat. Explore the Antivirus, Firewall & Deepfake Engine
Designed to patch vulnerabilities, remediate threats, and operate as a portable threat hunter capable of scanning external hosts or connected devices.
From remote offline communities (first aid, agriculture, survival) to webcam posture coaching and clinician triage, breadth of information remains educational and never replaces a licensed clinician.
Supported devices, installation requirements, and data-handling details belong with the app's release documentation.
WINTHOS AI ANTIVIRUS, FIREWALL & DEEPFAKE DETECTION ENGINE
Included free in Winthos AI Chat & Anti-Virus. The engine runs on-device and offline, combining behavioral and signature ML malware scanning with Real-Time Deepfake Detection (Audio, Video & Image Deepfake Verification) to detect synthetic voice clones, manipulated video calls, and AI-generated phishing media without mandatory daily database downloads.
Where the app scans files now
Real-time synthetic media & voice clone detection
Cross-OS, cross-CPU firewall & ML engine
No separate engine charge
ACROSS YOUR DIGITAL DAY
The engine is designed to detect malware, phishing, and real-time audio, video, and image deepfakes across files, calls, messages, email, and notifications. Today the app scans files on Android; full cross-platform deepfake call and message guards are in development.
Explore app availabilityThe app scans files on Android today.
Detects synthetic voice clones, manipulated video calls, and AI-generated phishing media in real time.
Designed to identify phishing signals and synthetic impersonation attempts across channels.
We have not published a detection rate or a false-positive rate, and the engine has not been independently tested. Coverage depends on the operating system, permissions, and integration. Nothing here establishes detection of every threat.
PUBLIC CHARITY INFRASTRUCTURE
Operated by 501(c)(3) public charity WIN'S VPN INC. with post-quantum ML-KEM-768 key exchange and Calculator Stealth Mode across Android, iOS, Windows, macOS, Linux, Smart TV, and Chrome Web Store.
WINTHOS HEALTH MINI / 256K CONTEXT
Built to bring information across medical specialties, psychology, movement, agriculture, and remote survival into a conversation you can understand.
Illustrative educational use: organize questions about prescriptions, lab results and clinical explanations for a qualified clinician. Imaging interpretation and persistent clinical memory are not verified features here.
Illustrative educational use: write down concerns and questions to discuss with a qualified mental health professional. Clinical counseling and continuous service availability are not established here.
Illustrative educational use: prepare questions about posture, exercise, or rehabilitation for a qualified professional. Live webcam posture coaching is in development for the app.
Illustrative educational use: organize dietary questions for a qualified professional. Meal-image diagnosis, deficiency detection and genetic personalization are not established here.
Illustrative educational use: organize observations and questions for a local agricultural adviser. Image-based soil or disease assessment and farm-record tracking are not verified features here.
First-aid education is an intended use, not a verified emergency-response capability. In an emergency, contact local emergency services immediately.
Winthos Health Mini is intended for educational access to medical information. Local-device execution and hardware requirements are pending verification. The model does not replace a licensed physician, psychologist, physical therapist, or emergency responder.
Emergency Notice: In a life-threatening emergency, call your local emergency number immediately. In the United States, call 911. For an emotional or mental health crisis in the United States, call or text 988. Outside the United States, contact your local crisis service. Diagnostic interpretation of medical imaging requires qualified clinical oversight. This notice is shown in English in every language.
How to evaluate health information · NIHWINTHOS SUPER
UP TO 1M CONTEXT · Clinical & U.S. Legal SpecialistMedical knowledge and specialized U.S. legal knowledge, brought together for professionals working with consequential information.
Winthos Super is intended for clinical information and U.S. legal workflows. Up to 1M context is planned for Dedicated Cloud GPU, API, and Max tiers. Casual web preview chats use a 64K to 128K token budget.
Through our legal intake properties (asklegalquestion.com, getafreelegalconsult.com, and applyforpermanentresidency.com), guided AI intake prepares paralegal-reviewed visual case summaries and $500 marriage-based permanent residency preparation packages (plus optional +$500 paralegal or attorney review).
“Clinical grade” describes Winthos's intended engineering standard, not regulatory authorization. Licensed physicians and licensed attorneys remain responsible for clinical judgment, legal analysis, citations, and court filings.
THE WINTHOS CYBER FAMILY
Winthos Cyber and Winthos Cyber Mini are engineered for software engineering, vulnerability remediation, and cybersecurity workflows alongside Real-Time Deepfake Detection.
WINTHOS CYBER MINI
A compact 1M-context cybersecurity model with a planned public weight release on Hugging Face once the Winthos community surpasses 10 million social-media followers.
The release plan is conditional. Weight availability and license terms will be published with the release.
WINTHOS CYBER
Flagship model positioned by Winthos for frontier-class software engineering, authorized security testing, and vulnerability remediation across our US and EU data centers.
Earlier unsupported Winthos Cyber benchmark scores have been withdrawn. Replacement results are TBA. See benchmark status and sourced references
PERFORMANCE, IN CONTEXT
Benchmark scores depend on the test, model configuration and evaluation method. Here is what the available evidence supports.
Sources checked · Auto-refreshed reference snapshot
Earlier unsupported Winthos scores have been withdrawn. Replacement results are TBA until complete evaluations are independently reviewed. No verified Winthos score or ranking is published here.
INDEPENDENT REFERENCE
Selected published model results, not a Winthos comparison or a complete leaderboard. Higher index scores are better. These are index points, not accuracy percentages. The index uses nine evaluations in four equally weighted categories. Read the methodology .
Other previously listed model/benchmark combinations: N/A (no verified result retained). Missing results are not estimated or counted as zero. Winthos replacement results are TBA.
Scroll across the table for configuration and source links.
| Model | Index score | Configuration | Evidence |
|---|---|---|---|
| Claude Fable 5.2 | 65 | Max | Source |
| Gemini 4 Carbon | 60 | High | Source |
| Claude Opus 5.5 | 59 | Max, default fallback | Source |
| GPT-6 Astra | 57 | Max | Source |
| GPT-6.1 Sol | 55 | Max | Source |
| Claude Sonnet 5.5 | 55 | Max, default fallback | Source |
| GPT-6.1 Terra | 51 | Max | Source |
| Gemini 4 Argon | 50 | High | Source |
| Muse Wonder Full | 50 | Max | Source |
| Kimi K3.1 Pro | 46 | As listed by AA; effort not specified | Source |
| Muse Wonder | 44 | Max | Source |
| Claude Haiku 5.5 | 36 | Max | Source |
| Muse Ember | 36 | Max | Source |
| Grok 4.7 | 35 | Xhigh | Source |
| Tencent Hy4 | 35 | As listed by AA; effort not specified | Source |
| Qwen3.8 Max (0902) | 34 | 0902 snapshot | Source |
SOURCE-SPECIFIC RESULTS
These results use their publishers’ own evaluation setups. They are not a common-harness ranking and are not directly comparable to Winthos.
HealthBench Professional
64.7length-adjusted points
Corrected in the September 22 system-card update. A rubric score, not a percentage of tasks passed.
Read OpenAI sourceInternal Research Debugging
78.05%OpenAI-reported score
OpenAI’s internal research-debugging evaluation. Its tasks and grading differ from the other results here.
Read OpenAI sourceTerminal-Bench 4.0
66.4%Anthropic-reported result
Reported at xhigh effort with production safeguards enabled; standard error is 2.6 points.
Read Anthropic sourceFrontierCode v1.1 (Main)
54.4%Anthropic-reported result
Evaluated with adaptive thinking at max effort and production safeguards enabled.
Read Anthropic sourceHealthBench Professional
58.1length-adjusted points
Evaluated by OpenAI using GPT-5.4 grading and Opus 5 fallback for provider refusals. This is not a Fable-only result.
Read OpenAI sourceTerminal-Bench 4.0
70.6%Anthropic-reported result
Anthropic’s evaluation setup. Results can change with the agent scaffold, tools and reasoning configuration.
Read Anthropic sourceA named model version, benchmark version and split, actual sample counts, scoring method, complete execution evidence and clear limitations. Repeated trials are not additional unique questions. A file checksum alone does not verify a result.
Earlier unsupported scores, receipt badges and the incorrectly labelled 57-benchmark average have been withdrawn. We do not claim Artificial Analysis has evaluated Winthos.
START WITH WINTHOS HEALTH MINI
Winthos Health Mini is designed to load in Winthos AI Chat & Anti-Virus alongside Real-Time Deepfake Detection (Audio, Video & Image Deepfake Verification). Official app and model download links are pending.
The app and model download URLs have not yet been supplied for this website. This panel will link directly to the official release when those destinations are available.
View Winthos Health Mini specificationsCAREERS · Q2 2027 COHORT · EU, USA & SINGAPORE
Accepting applications now for part-time and full-time openings starting Q2 2027 across the EU, USA, and Singapore (remote, hybrid, or 100% on-site for security-sensitive work) across Win Empire LLC (DBA Winthos AI) and WIN'S VPN INC.
Every employee holds a direct stake in our success and receives a percentage share of annual organizational income minus operating costs.
Even though we are an AI company, we pledge that we will never replace our human employees with AI.
Zero ATS or flawed AI resume filters. No automatic rejection for multi-page resumes, employment gaps, fresh-graduate status without internships, missing a tool brand, or outside commitments.
Zero micromanagement unless you ask for or need it. Finish weekly goals in 2 days without cutting corners and enjoy the rest of the week off, or bank months of paid vacation.
Advance specialist models, knowledge adapters, and inference systems toward our engineering team's end-of-2027 to end-of-2028 AGI research target (a forward roadmap goal, not an already established capability).
Partner with AI engineers on survey-backed human reasoning studies and solve difficult strategic challenges in ways others overlook.
Lead day-to-day defensive and offensive security engineering while integrating Winthos Cyber AI deployments for enterprise customers.
Dedicated security specialists and dual-role technical staff protecting offices, conference spaces, and proprietary IP against espionage or insider threats.
Guide regulatory compliance across local governments in the US, EU, and Singapore alongside internal corporate and public-charity governance.
Paid survey and reasoning-perspective research open without formal credentials so our models learn from diverse human thinking styles.
No degree gatekeeping: Formal degrees are not required; onboarding uses invite-only Winthos University modules. Stand out with real proof of capability that wows us, such as GitHub projects or community nominations. English proficiency is required (non-English names are always welcome) and verified via a recognized exam after an offer and before onboarding, with discretionary exceptions for minority or tribal languages lacking formal tests.
Principles & humanitarian R&D: Honesty, open-mindedness, and strict integrity around proprietary technology come first ("teach someone to fish and you feed them for a lifetime"). Beyond software and security, our R&D supports international peace initiatives and practical assistive technologies for blind and physically disabled people. Openings target Q2 2027 subject to accelerator/government grants and donor support; following, liking, and sharing our official social channels shows genuine human commitment rather than automated application spam.
LinkedIn applications only: We accept job applications exclusively through our official LinkedIn posting (linkedin.com/jobs/view/4488435920). We never recruit via WhatsApp or Telegram.
Verify email headers: Official candidate messages come only from our LinkedIn page or @winthos.ai and @winsvpn.tech addresses. Always check SPF and DKIM authentication and perform independent due diligence against spoofing.
TRY WINTHOS AI
Winthos SuperAn educational text preview of Winthos Super and Winthos Cyber, with thinking enabled and bounded replies. Answers may be wrong or incomplete. Do not enter sensitive health, legal or personal information.
Switch between Winthos Super and Winthos Cyber at any point. The most recent 10 messages carry across model switches.
The current daily allowance is shown in chat. Failed or canceled requests count once accepted. Replies are limited to 3,072 characters. Please ask one question at a time; pasting code projects or long files is not supported.
Sign in to see your account allowance. Please ask one question at a time; pasting code projects or long files is not supported.Signed in.
Sign-in did not complete. Please try again.
The sign-in request timed out. Please click the sign-in button again.
Sign-in is not available right now.
You are signed out.
Current daily limits and remaining attempts are shown in chat. Requests are paced and processed one at a time. One accepted message is one chat.
Your conversation starts here.
Paid priority and first-come ordering for free requests are planned. Queue ordering and congestion behavior have not yet been verified for this preview. Requests may wait, fail or need to be retried.
Messages are processed on Winthos servers in our United States (US) and European Union (EU) data centers. User-selectable US / EU datacenter routing is coming soon so you can pin prompts to your preferred jurisdiction. Chats are not saved to browser storage. Quota checks use a keyed hash (HMAC) of your IP address and UTC day. Our application does not persist raw IP addresses or chat transcripts. Logging by upstream infrastructure has not been verified.
If you sign in, we store a provider subject identifier, email address, display name, plan, credit balance and sign-in times for account access and usage accounting. We also store a hashed session identifier. If you request AYCE notifications, we record your interest and its timestamp. AYCE booking-request storage is described in Website privacy. We never receive your sign-in provider password. We do not store your messages or replies.
HIGH-FIDELITY REASONING
Ultra Mode pairs extended deep-reasoning verification with live service telemetry so hard security, coding, clinical, and legal prompts receive extra verification before delivery.
INSTANT ULTRA PASS
Get 5 high-compute Ultra Mode prompts with up to 4,096 tokens per prompt for $1. Built for testing deep-reasoning verification on complex security, coding, legal, or scientific questions without a monthly commitment.
Up to 4,096 tokens per prompt · US & EU data centersLIVE INFERENCE SPEED & LATENCY
Ultra Mode runs an extended deep-reasoning verification pass across our US and EU inference infrastructure to check calculations, citations, and edge cases before returning the final response. Human review remains essential for consequential decisions.
WINTHOS PLANS & GLOBAL ACCESS
Four monthly plans starting at $5/month for global and low-income affordability where $20/month subscriptions are out of reach. Keep your special rate while subscribed, or save another 20% with annual billing.
Half the planned standard price + double the usage = 4× usage per dollar.Compared with Winthos’s planned standard monthly price and allowance.
Built for global and low-income affordability worldwide
Planned standard price $10 / month
$5USD / month
12 special monthly payments $60
$48USD / year
Equivalent to $4 per month, billed annually.
Expanded daily allowance for regular builders
Planned standard price $20 / month
$10USD / month
12 special monthly payments $120
$96USD / year
Equivalent to $8 per month, billed annually.
For demanding engineering, clinical & legal workflows
Planned standard price $100 / month
$50USD / month
12 special monthly payments $600
$480USD / year
Equivalent to $40 per month, billed annually.
Our highest planned usage & 1M-context tier
Planned standard price $200 / month
$100USD / month
12 special monthly payments $1200
$960USD / year
Equivalent to $80 per month, billed annually.
THE FOUNDER’S RATE
Get 120 months of the $10 plan for one $500 payment. Your subscription ends after ten years, with no automatic renewal.
$500USD, paid once
About $4.17 per month over ten yearsCompared with $20 × 120 months = $2,400 at the planned standard monthly price.
Also saves $700 versus 120 payments at the $10 special monthly rate. The 2× usage bonus still ends on December 31, 2026.PAY-AS-YOU-GO AI CREDIT TOP-UPS
Light users can purchase non-expiring AI Credit Packs ($5, $15, $50) with zero monthly commitment and spend credits only when they chat (1 credit for Winthos Super Standard or Max, 2 credits for Super Ultra or Cyber Standard/Max, and 4 credits for Cyber Ultra). Heavy daily users get the best value from Winthos Global Access ($5/mo), Plus ($10/mo), Pro ($50/mo), or Max ($100/mo).
Starter top-up for occasional questions and second opinions
$5USD · one-time
Popular pay-per-use pack for regular research and code audits
$15USD · one-time
High-volume credit reserve for burst workloads without recurring billing
$50USD · one-time
Planned hourly access to a dedicated high-memory cloud GPU node. Booking and payment are not open yet.
$50USD per hour
Early access listing. Booking and payment are not open yet.
Sign in to request a time and to be notified when booking opens.
A time request does not reserve a node or take payment. Request at least 2 hours ahead, with a possible start up to 30 minutes earlier or later. Model and Ultra options are previews; requests record only a start time, length and time zone.
Open requests
Pick a day and a time.
hour
hours
Request received. Booking and payment are not open yet; we will confirm by email when they open.
You have already requested this time.
That time is too soon. Please pick a time at least 2 hours from now.
You already have 3 open requests. You can ask for another time after one of them has passed.
Please sign in first.
Too many attempts. Please wait a minute and try again.
Requests are not open right now.
Something went wrong. Please try again.
Thank you. We will notify you when booking opens.
Something went wrong. Please try again.
Credits per chat on paid plans: Winthos Super 1 for Standard or Max and 2 for Ultra; Winthos Cyber 2 for Standard or Max and 4 for Ultra. Winthos Super and Winthos Cyber are available in the preview chat today, and Ultra Mode can be activated with the $1 pass (5 short prompts).
Paid priority and first-come ordering for free requests are planned. Queue ordering and congestion behavior have not yet been verified for this preview. Requests may wait, fail or need to be retried.
Designed for usage comparable to similarly positioned ChatGPT, Claude, and Gemini plans. Exact model access, allowances, and fair-use limits will be published before checkout opens; competitor limits vary.
Prices are in USD. Monthly and annual plans renew until canceled, with special rates retained while active. The usage bonus ends on January 1, 2027. Standard prices are planned reference prices, not prices previously charged.
Annual prices are 20% below twelve special monthly payments: $60 becomes $48, $120 becomes $96, $600 becomes $480, and $1,200 becomes $960. The founder offer is a separate ten-year term.
Subscription checkout uses PayPal. Sign-in is required, and checkout is subject to availability. Service activates only after Winthos verifies payment and access.
WINTHOS API & IDE INTEGRATION
Paid accounts (active Winthos Global Access, Plus, Pro, Max, or Founder subscriptions, or accounts with an AI Credit Pack balance) can generate personal API keys below to connect Winthos Super, Winthos Cyber, and Winthos Ultra directly to Cursor, VS Code (Continue, Cline, Roo Code), Windsurf, or any OpenAI-compatible IDE and SDK across our US and EU data centers.
50% off-peak API discount between 8 PM and 8 AM Eastern Time once standalone metered billing opens, without requiring you to contribute your data to qualify.
Paid subscribers and AI Credit Pack holders can create up to 5 active IDE API keys today. Today, API chats spend AI Credits from your plan or credit pack (1 credit for Winthos Super Standard or Max, 2 for Super Ultra or Cyber Standard/Max, and 4 for Cyber Ultra); standalone per-token billing and function-calling tool schemas are not enabled yet. Before creating or using a developer API key, review the Terms and liability limits and Acceptable Use Policy. Keep keys private and use the models only within authorized work.
Sign in with a paid Winthos subscription or AI Credit Pack account to create, view, and revoke your personal wth_api_... bearer keys.
https://www.winthos.ai/api/v1winthos-super · winthos-cyber · winthos-ultraNew IDE API Key Created (copy now; for security, the full key is shown only once):
OpenAI Base URL: https://www.winthos.ai/api/v1
API Key: wth_api_YOUR_KEY
Models: winthos-super, winthos-cyber, winthos-ultra{
"title": "Winthos Cyber",
"provider": "openai",
"model": "winthos-cyber",
"apiBase": "https://www.winthos.ai/api/v1",
"apiKey": "wth_api_YOUR_KEY"
}curl -s https://www.winthos.ai/api/v1/chat/completions \
-H "Authorization: Bearer wth_api_YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"winthos-cyber","messages":[{"role":"user","content":"Audit this module."}]}'Planned API off-peak pricing: usage between 8 PM and 8 AM Eastern Time is 50% cheaper ($0.50 per 1M input tokens and $2.50 per 1M output tokens) once standalone token billing opens.
Winthos's planned standard API rates would total $6 for one million input tokens plus one million output tokens ($3 off-peak) once standalone token billing opens. Scan and remediate code vulnerabilities with Winthos Cyber first at low cost before running expensive final generalist passes.
ALREADY HAVE AN AI SUBSCRIPTION?
Put Winthos alongside the tools you already use. Check an answer against sources, or run an initial code security pass with Winthos Cyber before bringing in a high-cost generalist model.
Bring an answer from another AI and use Winthos to check it against primary sources. Follow the evidence and resolve disagreements before relying on the answer.
Hallucination-free answers are our goal. That claim awaits evaluation; a second model’s agreement is not proof that an answer is correct.
Our approach to evidenceUse Winthos Cyber for an initial security and bug review of your code, address its findings, then run a final audit with your preferred premium AI. Keep tests and human judgment in the process.
Our target is to identify 50% to 100% of the confirmed bugs and vulnerabilities that the final AI audit would find. Comparative evaluations are pending; this is a target, not a measured result or guaranteed coverage.
WHY WE'RE HERE
Technology is at its best when it gives people more agency. We believe AI should help a small team do ambitious work, help someone understand a difficult subject, and make knowledge less dependent on income.
Winthos means winning together. We want the gains from AI to create opportunity for people, including the people doing the work.
Explore our principlesHuman intelligence and artificial intelligence come together for a brighter tomorrow.
A symbol of guidance, hope, and new beginnings.
A brighter tomorrow for everyone.
A greater future. Together.
OPERATING ENTITIES & PUBLIC RECORDS
Winthos separates commercial software and AI operations from tax-exempt humanitarian public-charity programs, with verifiable Florida state filings and IRS public charity records.
Florida nonprofit corporation (Sunbiz N26000008302) recognized by the IRS as a tax-exempt 501(c)(3) public charity under section 509(a)(2) in 8 days (Form 1023 filed July 13, 2026; approved July 21, 2026). Holds an active Florida Department of Revenue Consumer's Certificate of Exemption (DR-14) valid through August 31, 2031.
Operates Win's VPN censorship-resistant internet access and public-interest educational initiatives. Inspect determination letters and filings at winsvpn.org/#proof.
Florida limited liability company (Florida Sunbiz L26000313506) operating Winthos AI commercial subscriptions, Developer API services, StartBusiness.tech small-business automation, and consumer software properties across US and EU data centers.
Commercial subscriptions purchase software access and do not constitute charitable donations; tax-deductible contributions go directly to WIN'S VPN INC.
Join Our Q2 2027 Cohort (EU, USA & Singapore): We practice Fordism profit-sharing, read 100% of applications by hand without ATS or AI resume filtering, and pledge never to replace human staff with AI.
INTERNATIONAL TAX-DEDUCTIBLE GIVING
Active U.S. 501(c)(3) / 509(a)(2) tax-deductible status today, paired with our four-jurisdiction international charity registration roadmap for direct local donor tax relief across the Netherlands, United Kingdom, France, and Canada.
| Jurisdiction | Structure / Portal | Donor Tax Relief Status |
|---|---|---|
| United States | WIN'S VPN INC. (Sunbiz N26000008302) | Active IRS 501(c)(3) / 509(a)(2) public charity |
| Netherlands | Foreign ANBI application via Belastingdienst (GlobalGiving guide, avoiding local KVK Stichting notary costs) | Planned direct EU Dutch tax deductibility on U.S. 501(c)(3) |
| United Kingdom | Charitable Incorporated Organisation via UK Charity Commission (model constitution) | Planned UK Gift Aid eligibility |
| France | Association loi 1901 via Service-Public + rescrit fiscal (domiciliation) | Planned 66% French income-tax receipt eligibility |
| Canada | Federal CNCA NFP via Corporations Canada (by-law builder) + CRA Registered Charity (CG-002 / CG-032; overcomes Article XXI U.S.-income cap) | Planned Canadian official RR donation receipts |
HELP MAKE ACCESS POSSIBLE
Support the work of making censorship-resistant internet access and practical technology available to more people. Contributions go directly to WIN'S VPN INC. via PayPal or cryptocurrency (BTC, DASH, XMR). A charitable donation does not activate a commercial Winthos AI subscription.
For IRS determination letters and Florida exemption records, visit Win's VPN's published records.
GLOBAL ACCESS PRICING
Commercial subscriptions start at $5/month so students, developers, and families worldwide are not locked out by $20 to $200 monthly bills.
Explore $5/month Global Access Explore all 9 ecosystem pillarsPROGRESS, WITH PEOPLE IN MIND
Technology does not exist in isolation. These are the values and workforce/economic policy proposals behind our work.
Many perspectives shape better decisions.
AI can inform the discussion.
People remain responsible for the choice.
Lower technology costs should give organizations room to invest in their people. We favor AI that expands human productivity, with thoughtful retraining and fair employment practices.
Our founder's experience in large organizations shaped concerns about contracting, unstable employment, and opaque outsourcing. We advocate clear compensation, transparent employment relationships, and accountability for the effect of automation on workers.
Public discussions about H-1B visas frequently confuse the statutory mechanics. The often-cited $60,000 figure is the annual wage at which workers of an H-1B dependent employer can be exempt from certain additional requirements (8 U.S.C. 1182(n)(3)(B); U.S. Department of Labor).
We support considering a $110,000 exemption threshold, alongside an assessment of its effects on wages, hiring and specialized roles.
The regular H-1B cap is 65,000, with an additional 20,000 exemptions for qualifying U.S. advanced-degree holders (8 U.S.C. 1184(g)). We favor transparent selection and enforcement against registration abuse; claims about processing delays and effects on domestic graduates require dated evidence.
Wage requirements · U.S. Department of LaborWe favor transparent staffing fees, clear compensation and accountable employment relationships. Workers should be able to understand how client payments relate to their compensation and employment terms.
We favor evaluating outsourcing arrangements for training quality, local systems knowledge, transparency and worker protections. Reform proposals should be assessed using documented labor-market evidence.
The EB-3 other workers category for qualifying nonseasonal permanent work has a statutory limit of 10,000 visas per fiscal year (8 U.S.C. 1153(b)(3)(B)). We support assessing labor needs using documented evidence.
One proposal we would support is establishing a five-year conditional legal status tied to verified agricultural and seasonal work. Completing five years of satisfactory service in essential food supply or seasonal infrastructure would earn permanent legal residency. The proposal should be assessed for its effects on labor availability, worker protections, farm finances and consumer prices.
We favor evaluating proposed datacenters against local electricity capacity, water availability, environmental effects and residential utility costs. Siting decisions should rely on evidence about the location and proposed facility. Winthos prioritizes energy-efficient, environmentally conscious computing infrastructure.
Data-center cooling and water efficiency · DOEWe want useful AI to reach people who cannot afford expensive tools. A verified project-cost summary is not yet published.
As a disabled creator overcoming technical barriers, the founder credits AI assistants with helping bridge technical knowledge gaps and physical limitations. Trust must be built through documented work, clear governance, and honest limits.
Community support helps fund public-interest access. The donation page identifies the recipient and its published terms.
AMBITION, WITH ACCOUNTABILITY
Trust should come from how a model performs and how openly its limits are explained.
Our training approach prioritizes authoritative material, including resources from NIST and NIH, with expert review as a goal for other sources. Dataset provenance and review coverage need to be documented, not assumed.
We aim to publish model versions, evaluation methods, hardware, and limitations alongside benchmark results. Coding, defensive security, medical, and legal evaluations must be assessed separately.
Lower inference costs, faster responses, and fewer unsupported answers are research priorities. Comparative results will follow measured, repeatable tests.
No AI is free from bias or mistakes. Our goal is balanced reasoning, clear uncertainty, and answers people can verify. Human judgment remains part of the process.
ENGINEERING PERSPECTIVE
Two centuries ago, suggesting that heavy machines would fly through the air invited ridicule; talking about landing on the moon sounded like sheer madness. Today, both are established engineering achievements.
We draw that exact parallel when applying disciplined training to cancer research and stubborn scientific challenges. We do not claim one model or one GPU cures cancer overnight. Pairing high-efficiency AI compute with the life-earned critical reasoning of human researchers makes hard discoveries achievable.
Through WIN'S VPN INC., an IRS 509(a)(2) public charity, public support expands open access: if you give someone a fish, you feed them for a day; teach them how to fish, and they eat for a lifetime.
Tax treatment depends on your situation. Ask a tax adviser. We do not give tax advice.
Winthos explores new ways of combining AI's ability to work across information with the judgment people develop through lived experience. The founder's long-term ambition includes contributing to difficult scientific problems, including cancer research and personalized health education.
These are long-term research aspirations. Neither AGI nor a medical breakthrough is established by a model description or benchmark percentage.
RESEARCH THAT INSPIRES US
RECURSIVE SELF-IMPROVEMENTThe Last AI Built by HumansMODEL SCALINGSOLAR 10.7B: Depth Up-ScalingINFERENCE EFFICIENCYKV Cache Compression ResearchIndependent research acknowledgments. These papers do not validate Winthos's performance or imply an endorsement.
HUMAN REVIEW MATTERS
AI output can be wrong or incomplete, including in security, code, health, and legal work. Have a qualified person verify consequential answers, test generated code, and obtain authorization before security testing. You are responsible for how you use the output and for harm caused by your use, subject to applicable law.
Read the Terms and liability limits and Acceptable Use Policy governing Winthos AI preview, subscription, credit-pack, and personal IDE API use.
BUILT WITH GRATITUDE
To the researchers who share their work, and the tools that help people build beyond their limitations: thank you.
Acknowledgments reflect the founder's account of the tools and research that supported this work. No partnership or endorsement is implied.
Yes. You can test Winthos Super, Winthos Cyber, and Ultra Mode right now on our live chat preview. Winthos Health Mini is designed to run on-device inside Winthos AI Chat & Anti-Virus, and Winthos Cyber Mini open weights are planned on Hugging Face at 10M+ community followers.
Winthos benchmark scores are withheld pending complete evaluation evidence. The benchmark section lists dated, sourced reference results for other models. It does not establish a Winthos ranking or a comparison under a common evaluation setup.
Paid accounts (active Winthos Global Access, Plus, Pro, Max, or Founder subscriptions, or accounts with a positive AI Credit Pack balance) can create, view, and revoke up to 5 personal IDE API keys directly in the Winthos API & IDE Integration section. Point your IDE (Cursor, Continue, Cline, Roo Code, Windsurf, or OpenAI SDK) to https://www.winthos.ai/api/v1 with model IDs winthos-super, winthos-cyber, or winthos-ultra.
No. This website does not accept document uploads. Its trial chat is text-only and processed server-side. Do not send sensitive health or legal information; use services with verified protections and appropriate professional oversight.
Charitable donations go to WIN'S VPN INC. (IRS 501(c)(3) / 509(a)(2) public charity). Cryptocurrency addresses match those published on winsvpn.org. A donation is separate from a commercial Winthos subscription; confirm any subscription amount, network, and payment reference with support first.