AI Providers

The built-in Hyper 30 model works out of the box — bring your own keys only if you want to.

HyperDevs ships with Hyper 30, a powerful cloud model for chat, coding, and generation. It's the default and needs no setup — just sign in. If you'd rather use your own account and billing, you can bring keys for Gemini, Claude, OpenAI, z.ai, OpenRouter, Groq, xAI (Grok), DeepSeek or Mistral, or run a local model with Ollama or LM Studio (no key needed), and switch providers per task. When you chat with a model that can see images, a screenshot of your game reaches it.

H

Hyper 30 Built-in · Default

No API key required

  • ✓ Works the moment you sign in — nothing to configure
  • ✓ Runs in the cloud, with no waiting in line: HyperDevs' proxy sends your messages through OpenRouter to providers that keep no data and don't train on it
  • ✓ Handles chat, Luau coding, and generation
  • ✓ Free to use every day with an account. A meter under the message box shows how much you've used, such as "Free today: 60% used", and when it resets (00:00 UTC, shown in your own time). A message you start with free use left finishes free, within limits
  • ✓ After that, keep going with credits if you turn it on: charged by how much the model reads and writes, about 5 credits a message and more for long builds, with a daily cap you set (100 unless you change it). Turn it off any time in Settings → Account
  • ✓ Replies can keep thinking for up to an hour. The chat shows "Reading your request…", then "Thinking · 3 min"
  • ✓ Memory scans on Hyper 30 use your free daily use and never credits: when it's gone, a scan stops, keeps its work and says when free use resets. Scans don't run during Hyper 30 maintenance

Available models: qwen3.8-27b (the only one)

Behind the name: a single dense Qwen3.8 27B served via the HyperDevs cloud proxy — not 30 separate models.

Find it under Settings → AI Providers → From HyperDevs. Signing in authenticates your HyperDevs account — this is a HyperDevs cloud credential, not a third-party API key. Messages use credits only once today's free use is gone, and only if you've chosen to keep going with credits. Credits also pay for image, 3D and animation generation (1 credit = $0.01), whichever provider you chat with. An image costs 2–12 credits and a 3D model 10–90, depending on the model.

Credits expire 6 months after they're granted. When some expire within 14 days, the balance in the title bar turns amber with an hourglass, and pointing at it shows how many expire and when. Settings → Account lists every expiry date, soonest first.

Bring Your Own Key

Optional. Add a key under Settings → AI Providers for any of these if you prefer your own account and pricing. A provider reads "Configured" only once its key is saved, and an edit you haven't saved says "Not saved".

The model menu scrolls. For providers with long lists (OpenAI, OpenRouter, Groq, Mistral) it has a search box and opens at the model you're using. If a model you saved has been retired, the app switches to the provider's recommended model.

G

Google Gemini

Affordable & fast

  • ✓ Very affordable
  • ✓ Fast response times
  • ✓ Great quality for coding & creative
  • ✓ Easy signup

Available models: Gemini 3.8 Flash (default), Gemini 3.6 Flash, Gemini 3.1 Pro (preview)

Get API Key →
C

Anthropic Claude

Complex coding

  • ✓ Excellent for Luau scripting
  • ✓ Strong reasoning abilities
  • ✓ Long context windows
  • ✓ More expensive but worth it

Available models: Claude Sonnet 5 (default), Claude Opus 5, Claude Haiku 4.5, Claude Fable 5.1

Note: a Claude Pro or Max subscription is separate and doesn't include API credit. You need an API key with credit from the Anthropic Console.

Get API Key →
O

OpenAI

All-around

  • ✓ Excellent across all tasks
  • ✓ Fast and reliable
  • ✓ Great creative capabilities
  • ✓ Custom base URL supported

Available models: GPT-5.6 Terra (default), GPT-5.6 Luna, GPT-5.6 Sol, GPT-6 Astra, plus whatever your account lists

Get API Key →
Z

z.ai (GLM)

Strong & cost-effective

  • ✓ GLM-5.3 (default), GLM-5.3 Flash, GLM-5.2, GLM-4.7
  • ✓ Capable reasoning and coding
  • ✓ Competitive pricing
  • ✓ Custom base URL supported

Available models: GLM-5.3 (default), GLM-5.3 Flash, GLM-5.2, GLM-4.7, plus your account's list

Base URL: the default is z.ai's Coding Plan endpoint. If you pay as you go with credits instead, change it in Settings to https://api.z.ai/api/paas/v4.

Get API Key →
OR

OpenRouter

Many vendors, one key

  • ✓ Models from every major vendor through one key

Available models: Claude Sonnet 5 (default), plus others such as GPT-5.6 Terra, Gemini 3.8 Flash, Grok 4.6, DeepSeek V4 Pro, Kimi K3 and Qwen3.8 27B

Get API Key →
G

Groq

Fast open models

  • ✓ Very fast inference for open models
  • ✓ Groq renames models often, so the list comes from your account

Available models: GPT OSS 120B (default), plus your account's list

Get API Key →
X

xAI (Grok)

Grok models

Available models: Grok 4.6 (default), Grok 4.5

Get API Key →
D

DeepSeek

DeepSeek V4 models

Available models: DeepSeek V4 Pro (default), DeepSeek V4.1 Flash

Get API Key →
M

Mistral

Mistral Medium & Codestral

Available models: Mistral Medium 3.5 (default), Codestral

Get API Key →

Local (no key)

Run entirely on your own machine — no account and no API key. Ollama downloads and serves models locally, and LM Studio serves the model you load in it.

🦙

Ollama

Free & offline (local)

  • ✓ Completely free
  • ✓ Runs locally (offline)
  • ✓ Private — nothing leaves your machine
  • ✓ The model you pick is the one that answers, even one you add while HyperDevs is open
  • ✓ If Ollama isn't running, or has no model loaded, the chat says so as soon as you pick it and tells you what to start

Available models: gemma4:26b (default), llama3.2, plus any model you've pulled

Setup: no key needed. Start it with ollama serve and pull a model. Default address: http://localhost:11434.

Download Ollama →
LM

LM Studio

No key (local)

  • ✓ Runs on your computer and keeps everything there
  • ✓ The model you pick is the one that answers, even one you add while HyperDevs is open
  • ✓ If its server isn't running, or has no model loaded, the chat says so as soon as you pick it and tells you what to start

Available models: whatever model LM Studio has loaded

Setup: no key needed. Load a model in LM Studio, then start its server (Developer tab). Default address: http://localhost:1234.

Download LM Studio →

💡 Tip: Start with Hyper 30 — no key, no setup. Switch to Claude or Gemini for specific coding tasks, or Ollama or LM Studio to keep everything on your computer. See Advanced for local model setup.

Documentation for HyperDevs v0.1.34 Report an issue