All modules

Every open AI model. One place.

There are thousands of powerful AI models anyone can run for free — Llama, Mistral, DeepSeek, Gemma and more. The catch is that installing them is fiddly. Models makes it two clicks: browse a catalogue, pick a model, click install. adris.tech figures out the right version for your laptop and runs it for you. Free.

00 What is Models? 01 One-click setup 02 Compare models 03 No GPU? Use the cloud 04 Curated collections

Run any AI model on your laptop — in two clicks, free.

Browse a catalogue · click install · adris.tech picks the right version for your machine

Hundreds of powerful AI models can be downloaded and used for free. The problem has always been the setup: figuring out which version fits your laptop's memory, installing extra software, configuring it, then wiring it into the tool you want to use.

Models hub removes all of that. You browse, pick, and click install. adris.tech auto-detects your RAM, downloads the right version, verifies it, and connects it to Krew and Coder automatically. The model runs on your own hardware — so you pay nothing per conversation.

DS
DeepSeek R1 · 7B
deepseek-ai
ReasoningCoding
Params7B
RAM4.5 GB
Speedfast
Run localCloud
LL
Llama 3.3 · 70B
meta
ChatWriting
Params70B
RAM40 GB
Speedmed
Run localCloud
MS
Mistral Small · 24B
mistralai
ChatQuick
Params24B
RAM14 GB
Speedfast
Run localCloud
QW
Qwen Coder · 32B
alibaba
Coding
Params32B
RAM20 GB
Speedmed
Run localCloud
SV
Sarvam · Indic
sarvam-ai
Indian langs
Params2B
RAM3 GB
Speedfast
Run localCloud
LL
Llama 3.1 · 405B
meta
Mesh frontier
Params405B
RAM230 GB
Speedmesh
Run on Mesh
The catalogue in the app — search by name or use case, filter by your laptop's RAM, install in one click.
Real story — Student researcher

Rashi is an engineering student in Tamil Nadu running an AI research project.

Her project needed her to test four different models on the same task. Paid AI services would have cost thousands of rupees in credits — and meant sending her research data to servers she didn't control.

She opened Models hub, found all four, and installed them in about 20 minutes — adris.tech picked the right version for her laptop each time. She ran her whole test suite locally. Total cost: ₹0. All data stayed on her laptop. She finished two weeks early.

Replaces Hugging Face (complex setup) Ollama CLI (manual) LM Studio Paid cloud AI APIs

No extra software. One click.

Open the Models tab → pick a model → click Download. The engine handles the rest.

No terminal, no external tools, no configuration. adris.tech detects your hardware, recommends a model that fits, downloads it into the app, and verifies it — then it's ready to use in Coder with one click.

// Models tab · DeepSeek R1 7B~90s to first token
Open the Models tab · browse the hub
Pick a model · click Download · watch the progress bar
Click Run · the model loads in Coder · start typing
→ detecting hardware… 16 GB RAM · 4 CPU cores
→ recommended: deepseek-r1-7b-q4 · 4.5 GB
→ downloading from adris.tech hub…
→ checksum verified · engine loading…
✓ ready · open Coder to start
It auto-detects your RAM and recommends a model that will actually run well on your machine.

Pick the right model — before you commit.

Run the same prompt across several models and see the outputs side by side

Stop guessing which model is best for your task. Download a few, type one prompt, and adris.tech runs them and shows the real outputs side by side — with real timing, no fake data.

Live comparison · inside the app

Download two or more models, type a prompt, and adris.tech runs them on your machine — one by one, or all at once if your RAM allows — showing the real outputs with timing.

01Download models from the hub above
02Open the Compare tab · select 2–4 models
03Type your prompt · hit Compare
04Real outputs · real timing · no fake data
Same prompt, same settings — the honest way to choose a model for your use case.

No GPU? Use the cloud.

Plug in your own OpenAI / Gemini / Claude key — at provider cost, zero markup

If your laptop can't run a big model, plug your own API key into Connect Apps. adris.tech routes Krew, Coder and Guard through your key at the provider's own price — no adris.tech markup. Your key, your bill, your data.

// Connect Apps → Add key → done
Step 1: Open adris.tech → Connect Apps
Step 2: Paste your OpenAI / Gemini / Claude key
Step 3: Hit Save & Test

adris.tech verifies the key with the provider,
then stores it in your OS keychain.
// ✓ connected · all AI features unlocked
{ "status": "connected",
  "provider": "openai",
  "model": "gpt-4o" }
ModelInput / 1kOutput / 1k
DeepSeek V3₹0.04₹0.12
Llama 3.3 70B₹0.06₹0.18
Sarvam-1 Indic₹0.02₹0.06
Claude Sonnet₹0.24₹1.20
Keys are stored encrypted in your OS keychain — never in plain text, never on our servers.

Curated by us. Updated regularly.

Short, opinionated lists for the situations you'll actually face

Browsing thousands of models is paralysis. So we publish short lists for real needs — "best models for Hindi", "runs on 8 GB RAM", "free coding model that beats the big ones". Curated, opinionated, current.

Collection · 12 models

Indian language stars

Best models for Hindi, Tamil, Telugu and Bengali — ranked on IndicGenBench.

Browse collection →
Collection · 8 models

Runs on 8 GB RAM

Smaller quantised models that fit a typical laptop without a discrete GPU.

Browse collection →
Collection · 6 models

Free coding heroes

Open models that match or beat closed ones on HumanEval — at zero cost.

Browse collection →
Collection · 10 models

Fast & tiny

Sub-3B models for quick tasks and lower-end machines. Very low latency.

Browse collection →
Opinionated short-lists, so you don't have to wade through the whole catalogue.

If you want real AI without a monthly bill, Models is the way in.

It fits best for:

People who benefit most
  • Students & researchers — unlimited local models at zero cost, with data that never leaves the laptop.
  • Developers — free coding models that plug straight into Coder.
  • Privacy-conscious users — run everything offline, nothing sent to a cloud.

Browse all models. Free, forever.

No signup needed to browse and download. Sign up only when you want hosted inference or to save collections.