Run JanitorAI with Nebula Block API: Fast, Free & Roleplay-Ready
JanitorAI has become popular for hosting AI-powered characters and chatbots. However, many of the APIs used with JanitorAI are now limited or shifting to paid-only access, making it harder for users to find reliable and free endpoints. A powerful alternative is Nebula Block’s serverless inference API — free for supported models and fully compatible with JanitorAI and any frontend that supports OpenAI-style endpoints.
This guide shows you how to set up Nebula Block as your backend for JanitorAI or similar interfaces.
Why Use Nebula Block with JanitorAI?
- Completely free for supported models — perfect for casual or high-volume use.
- No OpenAI key required — skip the hassle of registering external API keys.
- Supports top open-source models like DeepSeek, Meta LLaMA, Midnight, and Stheno.
- Includes select closed-source models for users seeking higher performance or unique capabilities.
- Serverless — no deployment needed, just plug and play via API.
- OpenAI-compatible endpoints — works seamlessly with JanitorAI and other similar frontends.
Whether you're roleplaying, experimenting, or building your own interface, Nebula Block offers flexibility for both developers and end-users.
Step-by-Step Setup Guide
1. Create Account
- Sign up Nebula Block account.
- No credit card needed for serverless usage.
2. Copy your API Key
- Navigate to API Keys.
- Copy the key and keep it safe.
3. Pick your chatbot on JanitorAI
For example, select Chat with Fia:
4. Set Nebula API in JanitorAI
- In JanitorAI or your compatible frontend, go to API settings.
- Choose Proxy
- Paste configurations from Nebula Block:
- Configuration Name: anything you want
- API Endpoint:
https://inference.nebulablock.com/v1/chat/completions - API Key: Copy your API Key which generated in step 2.
- Model: Choose from available ones like
mistralai/Mistral-Small-3.2-24B-Instruct-2506,Sao10K/L3-8B-Stheno-v3.2, etc.
5. Save & Start Chatting!
- That’s it. You can now interact with your characters through the Nebula Block-powered backend.
Benchmarking Roleplay Quality
In tests using the Fia character, various Nebula Block models were evaluated based on immersion, consistency, emotional depth, pacing, canon fidelity, and roleplay tension.
Here's benchmark and reviewer notes from the community:
| Model | Immersion | Consistency | Emotion | Pacing | Canon | Tension | Total | Strengths | Weaknesses | Reviewer Notes |
|---|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V3 | 8 | 9 | 8 | 9 | 8 | 9 | 51 | Sharp, grounded, emotionally stable | May lack warmth in lighter scenes | Excellent tone control, good for stoic RP |
| Midnight Rose | 7 | 7 | 7 | 8 | 7 | 8 | 44 | Strong presence, rich expression | Can over-emote in dramatic arcs | Great energy, but needs pacing balance |
| Mistral 24B | 8 | 9 | 9 | 8 | 8 | 9 | 51 | Natural flow, emotionally layered | Less distinct in unique characters | Calm, poetic — fits narrative RP |
| L3-8B Stheno | 7 | 8 | 7 | 7 | 7 | 7 | 43 | Clear logic, reliable responses | Feels more like prose than dialogue | Solid structure but lacks RP nuance |
| DS R1-0528 | 8 | 9 | 8 | 8 | 8 | 8 | 49 | Balanced, nuanced, consistent tone | Slightly dry in emotional arcs | Great subtlety and realism |
| Gemini 2.5 | 9 | 10 | 10 | 9 | 9 | 10 | 57 | Immersive, emotionally intelligent | May "over-perfect" some responses | Top-tier for immersive, serious roleplay |
Settings: Temperature:0.9, Max tokens: 0
- Canon: How well the model sticks to the established lore or backstory.
- Tension: Ability to create dramatic build-up in dialogue.
- Pacing: Rhythm of speech and emotional beats.
All models are tested via Nebula Block's serverless API, fully compatible with JanitorAI and other OpenAI-compatible front-ends.
Recommended Models
| Model | Best Use Case |
|---|---|
| DeepSeek-V3-0324 | Reasoning, coding assistants |
| meta-llama-3:8b-instruct | Natural conversation, general tasks |
| L3-8B-Stheno-v3.2 | Roleplay, creative writing, assistant-style dialogue |
| Mistral-Small-3.2-24B-Instruct-2506 | Long-context support, reduced generation errors |
| Midnight-Rose-70B-v2.0.3 | Roleplay, storytelling, benchmark-strong |
| DeepSeek-R1-0528 | Reasoning, math, and general-purpose tasks |
More models are regularly added. Check the Nebula Block playground for updates.
Tips
- You can switch models anytime depending on your character needs.
- Serverless mode may have rate limits. Upgrade to GPU instance if needed.
- No need to host your own server.
Community-Powered & Budget-Friendly
At Nebula Block, we believe AI should be accessible to everyone—not just big-budget teams. Our community-driven and cost-effective approach helps you build, experiment, and grow without barriers.
Community-Powered
- Collaborative Learning: Connect with a vibrant network of developers, roleplay creators, and AI enthusiasts. Share tips, use cases, and insights to grow together.
- Feedback-Driven Development: Your input shapes our roadmap. We actively integrate user feedback to improve services and meet real user needs.
- Resource Sharing: Access community-contributed configs, model tips, and API guides to streamline your workflow and experimentation.
Budget-Friendly
- Affordable GPU Access: Pay-as-you-go pricing on high-performance GPUs makes it easy for individuals and teams to scale affordably.
- Free Inference Endpoints: Test models like DeepSeek and Gemini for free—experiment without upfront costs.
- Reserved Instances: Lock in resources for steady workloads and enjoy lower rates with easier budget control.
By combining community knowledge and budget-friendly infrastructure, Nebula Block lets you focus on what matters—building innovative AI experiences.
Important: DeepSeek models is free in limited time.
Conclusion
Nebula Block offers a reliable, OpenAI-compatible, and most importantly free alternative to power JanitorAI and similar apps. With a fast-growing selection of powerful models and a plug-and-play API, it's an ideal solution for users who want flexibility without the cost.
Ready to Go Live?
Sign up and explore now.
🔍 Learn more: Visit our blog and documents for more insights or schedule a demo to optimize your search solutions.
📬 Get in touch: Join our Discord community or contact support for help.
Stay Connected
💻 Website: nebulablock.com
📖 Docs: docs.nebulablock.com
🐦 Twitter: @nebulablockdata
🐙 GitHub: Nebula-Block-Data
🎮 Discord: Join our Discord
✍️ Blog: Read our Blog
📚 Medium: Follow on Medium
🔗 LinkedIn: Connect on LinkedIn
▶️ YouTube: Subscribe on YouTube