List your company

Serverless Inference

Model hosting and inference · by Lilac

Frontier models with simple per-token pricing.

Visit website

What it does

Frontier models with simple per-token pricing. Use supported open models through an OpenAI-compatible API. No GPU reservation, no minimums, and no infrastructure to manage.

  • dedicated NVIDIA GPU infrastructure
  • OpenAI-compatible serverless inference
  • flexible options for committed capacity
Category
Model hosting and inference
Users
fast-moving startups
Public API
Available

Not yet disclosed: deployment options, platforms, languages and support channels.

Pricing

  • Free plan$0
  • Usage-based$0.05
    • Tokens (MiniMax M3 cache read)

Prices as published by Lilac, checked Oct 2026.

Security and compliance

As reported by Lilac. Ask for audit reports before you buy.

  • SOC 2In progress
  • API inputs and outputs are not stored at rest; exist only in memory during inference and deleted immediately afterward
  • No training on API inputs or outputs
  • No sharing of API data with third parties
  • No logging of prompts or completions; only metadata (request ID, token counts, sampling parameters) is logged for billing and debugging
  • Technical, organizational, and physical safeguards to protect information

Reviews

No reviews yet

Reviews come from people who use Serverless Inference at work. Share how your team uses it to help other buyers decide.

Write a review

Alternatives

  • MAGIC

    by Pipeshift

    Proprietary framework (Modular Architecture for GPU Inference Clusters) that compiles workload-specific inference pipelines, controlling layers from model to silicon including K...

  • Coding Platform (Kimchi)

    by Cast AI

    Enterprise AI coding and inference platform.

  • Red Hat AI Inference

    by Red Hat

    Optimize model performance with an integrated stack for fast, consistent, and cost-effective inference at scale.

  • Inference Endpoints

    by VESSL AI

    Coming soon. OpenAI & Anthropic-compatible APIs; pay only for the tokens you use.

  • QHexRT

    by RunAnywhere, Inc.

    An inference engine designed to run LLMs, VLMs, STT, TTS, and embeddings entirely on Qualcomm Hexagon NPUs.

  • Dedicated Model Inference

    by Together AI

    Deploy models on dedicated infrastructure.

Is this your product?

Claim it to keep details accurate and submit evidence for the next ranking review.

Claim Serverless Inference