Vivacious Cloud — Your training job, always on the cheapest GPU alive

From your machine

No dashboard required. Just your terminal.

Works the same in Windows, Mac, or Linux terminals. Copy, paste, run.

1 Install the CLI

Mac / Linux

curl -fsSL https://vivaciouscloud.com/install.sh | sh

Windows

iwr https://vivaciouscloud.com/install.ps1 -useb | iex

VIVACIOUS CLOUD — An Anirudha's Ambition

2 Log in from the CLI

terminal

vivacious login anirudha-s

VIVACIOUS CLOUD — An Anirudha's Ambition

3 Prepare your dataset locally

terminal

vivacious prepare <path>

Inspects files and computes SHA-256 manifest locally. No cloud transfer or GPU compute.

VIVACIOUS CLOUD — An Anirudha's Ambition

4 Permit the workload (Pre-Flight Check)

Step 4a (Full Fine-Tuning - Default)

vivacious permit anirudha-s ambition --model <model-id>

Runs full fine-tuning pre-flight check. Displays complete, transparent price upfront before deployment.

Step 4b (LoRA)

vivacious permit anirudha-s ambition --model <model-id> --method lora

LoRA parameter-efficient fine-tuning requiring less GPU VRAM.

Step 4c (QLoRA)

vivacious permit anirudha-s ambition --model <model-id> --method qlora

QLoRA quantized 4-bit fine-tuning for maximum cost savings.

VIVACIOUS CLOUD — An Anirudha's Ambition

5 Deploy the job

terminal

vivacious deploy anirudha-s

Uploads prepared dataset to Vivacious Storage and launches cheapest GPU instance across providers.

VIVACIOUS CLOUD — An Anirudha's Ambition

6 Monitor Balance & Exposure

terminal

vivacious balance

Displays server-authoritative live balance, total credit paid, and total spent.

VIVACIOUS CLOUD — An Anirudha's Ambition

7 Download Model Weights

terminal

vivacious download <job-id>

Downloads signed model weights tarball once job is completed and billing is settled.

VIVACIOUS CLOUD — An Anirudha's Ambition

⚡ Live Router Traffic Logs

router-traffic.log LIVE FEED

[16:14:00] Connecting to router feed...

[16:14:01] Established connection. Live sweep active.

1 Install the CLI

Mac / Linux

curl -fsSL https://vivaciouscloud.com/install.sh | sh

Windows

iwr https://vivaciouscloud.com/install.ps1 -useb | iex

VIVACIOUS CLOUD — An Anirudha's Ambition

2 Log in from the CLI

terminal

vivacious login anirudha-s

VIVACIOUS CLOUD — An Anirudha's Ambition

3 Prepare Dataset & Checkpoint Locally

Step 3a (Prepare New Dataset)

vivacious prepare dataset ./new_instructions.jsonl

Computes streaming SHA-256 fingerprint locally. Zero cloud transfer or GPU compute.

Step 3b (Stage Base Weights / Checkpoint)

vivacious prepare checkpoint ./my_custom_model.tar.gz

Scans safetensors allowlist & parses model config locally. Verifies zero pickle execution.

VIVACIOUS CLOUD — An Anirudha's Ambition

4 Permit the Retraining Workload

Full Fine-Tuning Retrain

vivacious permit anirudha-s ambition --model custom_model

Sizing engine sizes GPU requirements based on staged checkpoint architecture metadata.

LoRA Retrain on Custom Weights

vivacious permit anirudha-s ambition --model custom_model --method lora

Applies LoRA adapter fine-tuning directly on top of your existing custom base weights.

QLoRA (4-bit) Retrain

vivacious permit anirudha-s ambition --model custom_model --method qlora

Quantized 4-bit retraining for maximum memory efficiency and lowest spot rate.

VIVACIOUS CLOUD — An Anirudha's Ambition

5 Deploy the Retraining Job

terminal

vivacious deploy anirudha-s

Uploads both dataset and custom weights to Vivacious Storage and launches isolated GPU sandbox.

VIVACIOUS CLOUD — An Anirudha's Ambition

6 Monitor Telemetry & Balance

Track Live Job Progress

vivacious status <job-id>

Displays step progress, live training loss, VRAM consumption, and elapsed hours.

Check Balance & Spend

vivacious balance

Displays server-authoritative live balance, total credit paid, and total spent.

VIVACIOUS CLOUD — An Anirudha's Ambition

7 Download Retrained Weights

terminal

vivacious download <job-id>

Downloads signed model weights tarball once job is completed and billing is settled.

VIVACIOUS CLOUD — An Anirudha's Ambition

⚡ Live Router Traffic Logs

router-traffic.log LIVE FEED

[16:14:00] Connecting to router feed...

[16:14:01] Established connection. Live sweep active.

The fine print

Terms, Privacy & Governance

Transparent usage terms, strict non-custodial privacy, and automated billing guardrails designed to protect your compute and funds.

1. Service Scope & Autonomous Compute Routing

Intelligent Multi-Cloud Routing: Vivacious Cloud orchestrates containerized training workloads across verified GPU broker networks, matching your script with the most cost-effective hardware available.
Complete Intellectual Property Ownership: You retain 100% IP ownership and control of all model architectures, custom scripts, datasets, and output checkpoints submitted.
Non-Custodial Privacy: We strictly manage compute orchestration, telemetry, and checkpoint staging; our platform never inspects, alters, or retains private model data beyond your retention window.

2. Acceptable Use & Safe Execution

Valid Licensing: All submitted code, training datasets, and model architectures must belong to you or be appropriately licensed for commercial or open-source compute.
Hardware Sandbox Isolation: Workloads execute strictly within isolated hardware container sandboxes to guarantee multi-tenant security and system integrity.
Resilient Checkpointing: Checkpoint-enabled training scripts are encouraged; our systems can seamlessly migrate and resume supported workloads during remote provider disruptions.

3. Platform Security & Human Engineering Review

Automated Anomaly Detection (No Bot Bans): Automated security monitors strictly monitor network anomalies, cryptographic tokens, and compute quotas for safety. No account is ever suspended by an automated bot. Automated systems only flag unusual behavior for human review.
Human Engineering Verification: Every flagged event is personally investigated by our dedicated engineering team to prevent false positives and protect legitimate developers and students.
Fair Resolution & Account Protection: If malicious activity, payment fraud, or severe infrastructure exploitation is confirmed after engineering review, we act to protect the platform while maintaining transparent communication through our support channels.

1. Data We Maintain

Account & Metering Ledger: Basic profile details, encrypted transaction IDs, and cryptographic metering signatures necessary for verifiable billing.
Ephemeral Runtime Telemetry: Real-time loss metrics, iteration step numbers, and hardware utilization percentages emitted during active training runs.
Encrypted Model Output Archives: Completed model weight archives safely stored in high-performance cloud storage exclusively during your active retention period.

2. Data Confidentiality & Zero Training Guarantee

Zero Internal Training: We never inspect, read, sell, share, or train internal AI models on your private datasets, prompt completions, or model weights. Complete tenant isolation is guaranteed.
End-to-End Cryptographic Transport: All CLI communication, dataset uploads, and telemetry streams use authenticated, industry-standard TLS encryption.

3. Automated Retention & 7-Day Model Access

Immediate Dataset & Checkpoint Cleanup: Raw training datasets and intermediate staging checkpoints are automatically wiped immediately upon job completion or exit to protect your data privacy and eliminate unnecessary storage overhead.
Guaranteed 7-Day Model Retention: Your completed model weights remain securely stored and available for download via signed URLs for a full 7 days (168 hours) after job completion.
Persistent Access (Downloading Does Not Delete): Downloading your model does not delete it from our servers. Your weights remain accessible for the full 7-day window so you can re-download, verify checksums, or share with teammates at your convenience before automated purge.

1. Upfront Pre-Flight Price Estimates

Transparent Sizing: Figures calculated during pre-flight checks (duration, rate, and estimated total) are analytical estimates based on model parameter count, dataset volume, and real-time wholesale broker spot rates.
Pre-Execution Visibility: You review and authorize the total transparent cost upfront before any GPU instance is provisioned.

2. 10% Mid-Run Spot Surge Protection

Upfront Transparent Rate: The transparent total estimate is calculated and shown during your pre-flight permit check before any deployment begins.
Automated Budget Guardrail: If wholesale GPU broker spot rates fluctuate by more than 10% over your estimated rate mid-run, our system automatically pauses the workload into a safe zone. This protects your funds against unapproved cost spikes, allowing you to review telemetry and resume when ready.

3. Transparent Balance & Refund Policy

Real-Time GPU Hardware Allocation: Because GPU hardware instances are booked and consumed in real-time from upstream wholesale cloud providers, charges for elapsed compute cannot be refunded once delivered.
Non-Expiring Wallet Balance: Any unused deposited balance remains in your account indefinitely with zero expiration, ready for your next training run at any time.
Engineering Credit Assistance: If a training job fails due to an internal platform-side infrastructure issue or broker connection drop, our engineering team is here to assist. Contact support@vivaciouscloud.com with your job ID, and our team will review the telemetry logs and credit your balance promptly.

1. Six-Layer GPU Teardown Verification

Active Protection Against Runaway Bills: Unlike traditional cloud setups where forgotten instances generate massive unexpected bills, Vivacious Cloud deploys a 6-layer automated teardown architecture (job completion traps, heartbeat monitors, failover reconcilers, and broker inventory audits) guaranteeing that instances are cleanly terminated the moment training finishes.
Zero Lingering Charges: When your job concludes, all compute resources are de-provisioned immediately, ensuring zero post-run billing drift.

2. Dedicated Engineering Support & Reliability

Infrastructure Review Guarantee: Our engineering team actively monitors cluster health and node availability. In any rare event of platform infrastructure failure, our engineers review run logs to ensure fair credit resolution.
User Code Assistance: While users are responsible for training script parameters and hyperparameters, our support team is happy to provide guidance via support@vivaciouscloud.com to help you configure resumable checkpoints and optimize training efficiency.

Scroll container to read complete terms

Common questions

Frequently Asked Questions

Everything you need to know about our compute routing, data retention, pricing, and infrastructure policies.

Who is Vivacious Cloud built for? +

Vivacious Cloud is built for everyone who trains AI models: from students and academic researchers looking for affordable compute without complex cloud configurations, to solo developers launching custom models, to fast-growing startups and enterprise teams running heavy production workloads. Our distributed orchestrator automatically provisions the exact GPU architecture you need with zero DevOps management.

How long are my model weights stored? Will downloading them delete them? +

Downloading your model does not delete it. Your trained model weights remain securely stored and available for download for a full 7 days (168 hours) after job completion. You can download your weights multiple times, verify them locally, or share download access with team members at any time during this 7-day window. Raw datasets and intermediate checkpoints are wiped immediately upon job completion to maintain privacy and eliminate storage costs.

How does transparent pricing and the 10% price protection policy work? +

Before launching any job, running vivacious permit calculates and displays the transparent total price upfront based on your model size and current wholesale broker rates — there are no hidden surprises. While training, our 10% Price Protection Policy continuously monitors live spot rates. If market spot prices fluctuate more than 10% above your estimated quote mid-run, the orchestrator automatically pauses your workload into a safe zone, preventing unexpected charges and keeping your budget safe.

Will an automated bot ever suspend my account? +

No. Automated bots never suspend accounts. Our automated security systems strictly perform anomaly detection and only flag unusual activity for safety. Every flagged event is personally investigated and verified by our real human engineering team to prevent false positives and protect legitimate developers and students.

What is your refund and credit policy? +

Because GPU compute instances are allocated and consumed in real-time from upstream wholesale cloud brokers, funds consumed for active compute cannot be refunded. However, your prepaid account balance never expires and can be used whenever you are ready. Furthermore, if a training run encounters a verified platform-side infrastructure fault, our engineering team will review the telemetry and credit your balance promptly.

How does the 6-layer GPU teardown protect me from runaway bills? +

A major frustration with traditional cloud providers is zombie GPU instances that keep billing after a script finishes or crashes. Vivacious Cloud features a multi-layer automated watchdog across job exit codes, failure traps, heartbeat monitors, and cloud broker inventory reconciliation to guarantee that all remote GPU instances are cleanly shut down the moment your training concludes or pauses.

How can I reach your team for technical support or guidance? +

Our dedicated engineering team is readily available:

• Technical Support & Workloads: Email support@vivaciouscloud.com for script configuration, training session debugging, and model setup assistance.

• General Enquiries & Billing: Email help@vivaciouscloud.com for onboarding questions, account inquiries, and billing support.

• Community & Live Chat: Join our Telegram community at @iEverYours.

Stop renting one cloud. Start renting the cheapest one.

Built for students, solo developers, and enterprise scale.

Zero-config deployment

Real-time price arbitrage

Mid-job provider switching

Zero-egress storage

Three steps. That's the whole onboarding.

Make your account

Add your account details

Run the terminal commands

No dashboard required. Just your terminal.

Transparent compute. Zero subscriptions.

Upfront Transparency

Spot Surge Guardrail

Zero Idle Cost

Terms, Privacy & Governance

1. Service Scope & Autonomous Compute Routing

2. Acceptable Use & Safe Execution

3. Platform Security & Human Engineering Review

1. Data We Maintain

2. Data Confidentiality & Zero Training Guarantee

3. Automated Retention & 7-Day Model Access

1. Upfront Pre-Flight Price Estimates

2. 10% Mid-Run Spot Surge Protection

3. Transparent Balance & Refund Policy

1. Six-Layer GPU Teardown Verification

2. Dedicated Engineering Support & Reliability

Frequently Asked Questions

Your next training run could already be on a cheaper GPU.