Start
Free
A restrained entry into AGI 01, LucidArc, and LucidCode.
€0forever
Get started- Limited AGI 01 Swift
- Limited AGI 01 Frontier
- LucidArc essentials
- LucidCode CLI included
LucidQueryAGI 01 · Swift and Frontier
Frontier AI,
engineered in Europe.
AGI 01 takes on the work a team actually has: research against live sources, code written and reviewed, documents and images read, decisions backed by something you can check. In chat, on your desktop, and behind an OpenAI-compatible API — with European data handling throughout.
The familytwo models · one family
AGI 01 Swiftgeneral · fast
AGI 01 Frontierflagship · max
- AGI 01
- model family
- Live
- web search
- Code
- native coding
- Photo
- to location
Models
Two models. One family.
Swift and Frontier are separate models, not one model at two speeds. Swift handles the everyday flow; Frontier goes deeper. Both search live information, write code, reason through tasks, and understand the world from text or images.
AGI 01 Swift
General · Fast
The fast everyday model for search, writing, coding, image understanding, and lightweight agents. It moves quickly and stays useful without overthinking the task.
950
LucidQuotient
98.2%
MATH-500
AGI 01 Frontier
Flagship · Max
The strongest all-purpose model for deeper reasoning, code review, research, long tasks, and real-world context. Use it when the answer needs more judgment.
1125
LucidQuotient
xhigh
Reasoning
Two configurations
Benchmarks
Measured, not claimed.
One composite score and the public tests, shown in full rather than cherry-picked.
Composite
One number across every axis
LucidQuotient folds reasoning, coding, vision, deduction, research, and prediction into a single composite. AGI-01 Frontier-preview reaches 1125 — past Claude Opus 4.8 and closing on the frontier, with Swift close behind.
1125
AGI-01 Frontier-preview
950
AGI-01 Swift
Public · zero-shot
Swift, on the standard tests
At pass@1, AGI-01 Swift posts 94.7% on HumanEval, 84.3% on GPQA Diamond, and 98.2% on MATH-500 — a clear jump over the earlier Nexus Coder and RF1 generations.
94.7%
HumanEval
84.3%
GPQA Diamond
98.2%
MATH-500
* LucidQuotient is LucidQuery's own composite benchmark, not third-party verified. Competing models are evaluated at maximum reasoning effort. Frontier-preview is an early preview — its score is not final.
Composite + public
The thesis
Frontier work shouldn't
need a data centre.
Parameter count stopped being an honest proxy for capability. Our post-training, with the World Engine driving the model at run time, already puts AGI 01 ahead of models several times its size. The bet we're making next is where that leads: intelligence small enough to run on hardware you already own, with your data never leaving it.
4×
the size of the models AGI 01 goes toe to toe with.
Post-training plus World Engine orchestration at run time. Home-scale deployment is the roadmap, not a shipped feature.
Engineered in France
Platform
Search, code, see, and act from one place.
The work feels simple because the complexity sits inside AGI 01: live information, image understanding, coding, and multi-step execution in a single model experience.
01Core
A model that reaches out
AGI 01 can search current sources, inspect context, write code, and continue across steps without making you wire together a stack of tools.
02Live
Real-time search
Ask about what is happening now and get answers grounded in live information, not stale memory.
03Vision
Photo to place
Drop in a picture and AGI 01 can infer where it was taken from visual clues, landmarks, language, light, and geography.
04Code
Native coding
Generate, review, explain, and refactor software with a model that understands product intent as much as syntax.
05Agents
Agentic workflows
Break down goals, keep context, and move through research, planning, drafting, and implementation in one flow.
06Adaptive
Right depth automatically
Swift answers quickly. Frontier thinks longer when the work needs more care, judgment, or synthesis.
07Trust
European engineering
A premium AI stack built with privacy, product quality, and operational trust at the center.
7 capabilities
Vision
A single photo. The exact place.
AGI-01 reads what a human can't at a glance — the quality of the light, the architecture, the script on a street sign, the slope of the horizon — and pinpoints where an image was taken. No GPS tags. No metadata. Just the picture.
Previously known as StrikeLoc, the system has been rebuilt and integrated directly into AGI-01, bringing photo geolocation into the same reasoning flow as search, coding, and analysis.
- 96%
- median accuracy
- < 10s
- to a confident answer
- 1 image
- no metadata needed
96% match
Santorini, Greece
36.3932° N, 25.4615° E
Click to collapse
Photo → coordinates
API
A drop-in unified API.
Fully compatible with the OpenAI interface. Keep your code, change one base URL, and reach AGI 01 Swift or Frontier.
from openai import OpenAI client = OpenAI( api_key="lq_live_...", base_url="https://api.lucidquery.com/v1",) stream = client.chat.completions.create( model="lucidquery-agi-01-frontier", messages=[{"role": "user", "content": "Explain quantum error correction."}], stream=True,) for chunk in stream: print(chunk.choices[0].delta.content or "", end="")01
OpenAI compatible
Point your existing SDK at our endpoint. Zero rewrites.
02
Swift or Frontier
Choose the speed you need with one model string.
03
Streaming
Responses arrive as they are written, ready for product interfaces.
04
Enterprise security
Scoped keys, audit trails, European data handling.
OpenAI-compatible
LucidCode
Plan it. Build it, in your terminal.
LucidCode is an agentic AI coding agent. It reads your codebase, plans an approach, then writes and edits code as reviewable diffs, running entirely from your command line and powered by AGI 01.
- A real agentic tool loop: reads, plans, edits, runs
- Plan and Build modes, one Tab apart
- Powered by LucidQuery AGI 01
npm · terminal agent
LucidArc
AGI 01, one keystroke away.
LucidArc is the command layer for your desktop. Search, compute, translate, summarize, browse, and ask AGI 01 without leaving the window you are already in.
- Search files and the web instantly
- Calculate, convert, command
- Call AGI 01 from anywhere
LucidArc — Presentation
Windows · command bar
Pricing
Everything you need. One plan.
Use AGI 01 across chat, API, and LucidArc with clear tiers and no feature maze.
Save ~20% with annual billing
Best value
Premium
The complete Lucid experience for everyday work.
€10.99/ month
- Everything in Free
- Near-infinite AGI 01 usage
- Frontier's full power
- LucidArc premium included
- LucidCode with premium models
- Priority speed for larger tasks
Teams
Enterprise
Control, scale, and assurance for organizations.
Custom
Talk to us- Everything in Premium
- Team controls
- API scale and SLAs
- Dedicated support
One plan · no maze
Put AGI 01 to work.
The report due Friday. The numbers buried in a spreadsheet. The research that's been sitting in your tabs. Bring it to AGI 01 — it searches, reasons, and follows through in one place.