How much does Lilac cost for batch GPU jobs?

Batch GPU container jobs on Lilac are priced per second: $1.00 per hour for H100s and $1.50 per hour for H200s. There are no minimums or contracts—you pay only for the compute seconds used.

Does Lilac offer a free trial or free credits?

Lilac does not offer a free trial. However, it has a subscription credit model where credits can stretch up to 12x when using idle supply. New users receive pay-per-token inference access with no upfront cost.

What are the best alternatives to Lilac for idle GPU monetization?

Alternatives to Lilac include RunPod, Vast.ai, and SaladCloud for decentralized GPU compute. Unlike those, Lilac focuses on monetizing existing enterprise clusters via a Kubernetes operator, with supplier payout of 70% revenue and no data leaving infrastructure.

What are the limitations of Lilac?

Lilac's key limitations include no dedicated GPU guarantees (availability depends on idle supply), limited model selection (MiniMax, Kimi, GLM, Gemma 4), and no managed training service. Also, batch jobs only support container workloads, not interactive sessions.

What integrations does Lilac support?

Lilac provides an OpenAI SDK-compatible API for inference. It integrates via a Kubernetes operator for supplier nodes. As of now, no other explicit integrations are documented.

Is Lilac easy to set up and use for inference?

For consumers, Lilac is easy to use: inference uses an OpenAI-compatible API, so swapping from standard OpenAI endpoints requires minimal code changes. Set up involves obtaining an API key and pointing your client to Lilac's endpoint.

Is Lilac good for running large language models with long context windows?

Yes, Lilac supports context windows up to 1 million tokens on MiniMax M3, making it suitable for long-document analysis or extended conversation tasks. Other models may have shorter context limits. Pricing is pay-per-token with cache discounts for repeated context.

How does Lilac compare to standard cloud GPU providers like AWS or Lambda?

Lilac offers 30% cheaper inference tokens and 60% cheaper batch jobs vs. on-demand clouds by leveraging idle supply. However, unlike AWS or Lambda, Lilac does not guarantee dedicated availability or have a broad model catalog. It's best for cost-sensitive, non-critical workloads.

Can I monetize my existing GPU cluster with Lilac?

Yes, organizations with idle GPU clusters can monetize spare capacity by installing Lilac's Kubernetes operator on their cluster. GPUs never leave your infrastructure, and you earn 70% of the revenue from inference and batch jobs. No long-term contracts or minimums required.

Lilac

Q: Who is Lilac best for?

Lilac is best for organizations with idle GPU clusters wanting to monetize spare capacity, AI developers seeking low-cost inference on frontier models, teams running batch GPU jobs on a budget, and startups needing flexible cluster reservations without long-term contracts.

Q: Who should avoid using Lilac?

Lilac is not for users needing guaranteed dedicated GPU availability, those requiring a wide selection of proprietary models, enterprises that cannot allow third-party workloads on their infrastructure, or users looking for a fully managed model training service.

Paid

Monetize idle GPUs or run inference/batch compute cheaply on a decentralized network.

By Tanmay Verma, Founder · Last verified 03 Jul 2026

0 views

Added 7d ago

77/100Safe Bet

Visit Website

In short

Lilac — Monetize idle GPUs or run inference/batch compute cheaply on a decentralized network. Best for Organizations with idle GPU clusters wanting to monetize spare capacity, AI developers seeking low-cost inference on frontier models, Teams running batch GPU jobs on a budget. Plans from $1/mo.

Compared withvs Voyage Ai vs Spider Cloud vs Temporal Ai

Is Lilac actually worth it?

Live

See what real users actually say. We scan live discussions, reviews and complaints across the web and hand you an honest verdict — in under a minute.

3 free scans · no card needed · downloadable report

Run a free scan

Editorial Verdict

Best for

Organizations with idle GPU clusters wanting to monetize spare capacityAI developers seeking low-cost inference on frontier modelsTeams running batch GPU jobs on a budgetStartups needing flexible cluster reservations without long-term contracts

Not ideal for

Users needing guaranteed dedicated GPU availabilityThose requiring a wide selection of proprietary modelsEnterprises that cannot allow third-party workloads on their infrastructureUsers looking for a fully managed model training service

A smart two-sided marketplace for GPU supply and demand. Great for budget-conscious AI teams and cluster owners with spare capacity, but not suitable for workloads needing guaranteed availability or broad proprietary model access.

Last verified: July 2026

What's new in Lilac

Checked 7 days ago

Across the latest 8 updates: 1 feature update, 3 launches and 4 news mentions.

NewsBlog·May 7Newest

How to keep frontier open weights viable

Lilac argues that commercial rights help keep frontier models open, supporting open-weight licensing.

LaunchBlog·May 7Newest

We're partnering with MiniMax to bring M2.7 to Lilac

Lilac partners with MiniMax to offer commercially licensed M2.7 access on its idle-GPU network.

LaunchBlog·May 1

Kimi K2.6 is live on Lilac

Kimi K2.6 joins Lilac with OpenAI-compatible chat, 262K context, and cache-read pricing.

FeatureBlog·Apr 27

Cache read pricing is now live on Lilac

Lilac introduces cache-read pricing for repeated context, reducing long-context and agent workload costs.

NewsBlog·Apr 8

GLM 5.1 Inference Benchmark

Lilac benchmarks GLM 5.1 endpoint against other providers, claiming competitive throughput at lowest price.

LaunchBlog·Apr 8

Lilac is now self-serve — plus GLM 5.1 and Gemma 4 are live

Lilac removes waitlist for self-service API keys; adds GLM 5.1 and Gemma 4 models.

NewsBlog·Mar 25

How Idle GPUs Make Cheap Inference Possible

Lilac explains how idle enterprise GPUs enable low-cost Kimi K2.6 inference via shared endpoints.

NewsBlog·Mar 23

GPU Inference API Pricing Compared

Lilac publishes pricing comparison across GPU inference providers, highlighting idle-GPU cost advantage.

Viability Score

77/100

Safe Bet

How likely is Lilac to still be operational in 12 months? Based on 4 signals — momentum (how recently it shipped), wrapper dependency, revenue model, and web presence.

momentum

funding runway

website health

wrapper dependency

100

Last calculated: July 2026

How we score →

Key Features

Pay-per-token inference on open frontier models
Subscription credits with up to 12x idle-supply value
Batch container jobs priced per second on idle GPUs
Cluster reservation brokered from neo-cloud partners
Supplier monetization via Kubernetes operator
OpenAI SDK compatible API for inference
Supports MiniMax M2.7, M3, Kimi K2.6, GLM 5.1, GLM 5.2, Gemma 4 31B
Quantization support: FP8, INT4, NVFP4
Context windows up to 1M tokens (MiniMax M3)
Cache read pricing for repeated context
No contracts or minimums for inference and batch
GPUs never leave supplier infrastructure
Per-cluster reporting for suppliers
Private beta for batch jobs
Supplier onboarding via demo and waitlist

About Lilac

PaidIntermediateAPI availableAPI · CLI

Lilac is a network of idle GPUs that lets organizations monetize spare compute capacity and AI teams run inference, batch jobs, and cluster reservations at reduced costs. Instead of building a new data center, Lilac taps into existing GPU clusters that sit 30–50% idle, surfacing that capacity as a buyable resource. Suppliers install a Kubernetes operator on their cluster—GPUs never leave their infrastructure—and earn 70% of revenue. Consumers pay per token for frontier models (MiniMax M2.7, M3, Kimi K2.6, GLM 5.1/5.2, Gemma 4 31B) via an OpenAI-compatible API, or subscribe for credits that stretch up to 12x on idle supply. Batch container jobs run on H100s at $1/hr and H200s at $1.50/hr, priced per second. Cluster reservations are brokered from neo-cloud partners with estimated rates starting at ~$2/hr for H100s. The platform supports quantizations FP8, INT4, NVFP4, and context windows up to 1M tokens (MiniMax M3). Backed by Y Combinator, it’s live with customers like Z.ai, Osmosis, and Saturn Cloud. Where Lilac differentiates is its focus on idle supply from existing clusters—leading to 30% cheaper inference tokens and 60% cheaper batch jobs versus on-demand clouds, with no contracts or minimums. However, it doesn’t offer guaranteed dedicated capacity or a wide proprietary model selection, and supplier onboarding is by demo and waitlist.

Behind the Verdict

Lilac solves a real problem: most GPU clusters sit partially idle. If you own a cluster, Lilac’s Kubernetes operator makes it easy to monetize that downtime without sacrificing priority for your own jobs. The 70% revenue share is better than most resale arrangements. For consumers, pay-per-token pricing on frontier models like MiniMax M3 and Kimi K2.6 is genuinely cheap—30% below typical on-demand rates—and the OpenAI-compatible API means zero code changes. The subscription credits (Basic $10/mo, Pro $30/mo, Max $100/mo) can be a steal when idle supply is high, but value depends on network activity. Batch jobs at $1/hr for H100s are competitive with spot instances, and pricing per second avoids waste. The biggest caveat: availability is not guaranteed. If the network is busy, you may face queue delays or higher cache-read rates. Supplier onboarding is still manual (demo + waitlist), so it’s not plug-and-play yet. Compared to other decentralized GPU networks like Akash or RunPod, Lilac differentiates with its focus on warm idle GPUs (not cold spot instances), yielding lower latency for inference. But it lacks the breadth of model selection you’d find on Together AI or Replicate. Best for: teams already using Kubernetes who want to monetize spare capacity, or AI developers who can tolerate spot-style compute for cost savings. Not ideal for production workloads requiring SLA-backed uptime or a wide range of proprietary models.

Researching Lilac? Get your full AI stack in 60 seconds.

Free, no signup — tell us your goal and get tools matched to your budget & existing stack.

Use Cases

Run inference on open frontier models with pay-per-token pricing, compatible with OpenAI SDK.
Submit containerized batch jobs to run on idle GPUs at $1-$1.50 per hour.
Monetize underutilized GPU clusters by installing the Lilac Kubernetes operator and earning 70% revenue share.
Reserve dedicated clusters via brokered quotes from neo-cloud partners for longer-term projects.
Use subscription credits that multiply in value when GPU supply is idle, stretching your budget.

Models Under the Hood

MiniMax M2.7MiniMax M3Kimi K2.6GLM 5.1GLM 5.2Gemma 4 (31B)

Limitations

Batch jobs are in private beta, limiting availability.
Supplier onboarding requires a demo and waitlist, so immediate setup is not possible.
Cluster pricing is estimated and varies by provider, not guaranteed.
The platform focuses on open models, lacking many proprietary options.

12-month cost

Project the real annual outlay, including the implied monthly cost when only an annual tier is published.

Plan

Annual total

$120

Over 12 months

Effective monthly

$10

Billed monthly

Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.

Integrations

Kubernetes

Resources & Guides

Frequently Asked Questions

Featured Head-to-Head Comparisons

Lilac vs Voyage Ai

Lilac vs Spider Cloud

Lilac vs Temporal Ai

Popular in Developer Infrastructure

Temporal AI

Durable execution platform for reliable AI agents and workflows.

FreemiumTry

Spider Cloud

Fast web crawling, scraping, and search API for AI agents

FreemiumTry

Voyage AI

Domain-specialized embedding models and rerankers for enterprise RAG pipelines.

Contact SalesTry

Used Lilac? Help shape our editorial sentiment research.

Editorial Verdict

Best for

Not ideal for

Last verified: July 2026

Key Features

Pay-per-token inference on open frontier models

Subscription credits with up to 12x idle-supply value

Batch container jobs priced per second on idle GPUs

Cluster reservation brokered from neo-cloud partners

Supplier monetization via Kubernetes operator

OpenAI SDK compatible API for inference

Supports MiniMax M2.7, M3, Kimi K2.6, GLM 5.1, GLM 5.2, Gemma 4 31B

Quantization support: FP8, INT4, NVFP4

Context windows up to 1M tokens (MiniMax M3)

Cache read pricing for repeated context

No contracts or minimums for inference and batch

GPUs never leave supplier infrastructure

Per-cluster reporting for suppliers

Private beta for batch jobs

Supplier onboarding via demo and waitlist