AI Research by Clatter Labs

Research that turns AI capability into dependable software.

We study models, inference and agentic systems, and publish what we learn, so builders can trust what they ship.

Hands cradling a lotus that blooms into a network of connected nodes

Our focus

The difficult part is not calling a model. It is designing everything around it: the code, context, tools, evaluation, deployment, and feedback loops. That is the layer we study.

Research areas

Three questions we keep coming back to.

Illustration for agentic systems and coding agents

01

Agentic systems

How coding agents plan, use tools and recover from mistakes: the work behind Line CLI and the Cosmos Agent API.

Illustration for evaluation, honesty and robustness

02

Evaluation & honesty

Hallucination traps, unverifiable claims, sycophancy, pressure and injected instructions, tested in English and Egyptian Arabic.

Illustration for inference and efficiency

03

Inference & efficiency

Tokens, latency and cost per task measured as first-class results, not footnotes.

Approach

How the lab works.

  1. 01

    Software first

    Every project starts with a real engineering problem.

  2. 02

    AI as a material

    Models are system components with inputs, tools, evaluation and operational boundaries.

  3. 03

    Open by default

    We publish research, writing and much of our code.

Illustration for open research and publications
  1. 01Research
  2. 02Prototype
  3. 03Ship
  4. 04Iterate

The dotted line loops each iteration back into research.

Benchmark 3 · November 6, 2026

SuperCosmos 2.0, measured against 1.1.

A first, small run. 2.0 is within one problem of 1.1 on coding and one to two cases behind on honesty and purity, while using more tokens and time per task. Small samples: treat it as a first comparison, not a ranking.

Max tier shown, plus Flash. Pass rates by suite; cost in US dollars.
Model Honesty n=22 Purity n=22 HumanEval+ n=19 MBPP+ n=15 Cost / 1,000 tasks
SuperCosmos 2.0 Flash 90.9% 95.5% 89.5% 86.7% $3.29
SuperCosmos 2.0 Max 86.4% 95.5% 94.7% 86.7% $5.55
SuperCosmos 1.1 Flash 100.0% 100.0% 94.7% 86.7% $1.76
SuperCosmos 1.1 Max 95.5% 100.0% 94.7% 86.7% $2.66

From research to products

What the research ships as.

01

Beta

Line CLI

A very fast coding agent for your terminal, written in Rust. Slash commands, auto routing across the SuperCosmos models, skills, plugins and cloud delegation.

Explore Line CLI

02

Alpha

Cosmos Agent API

An AI agent API for coding agents. Start a run, stream every step and get the result, with scoped keys.

Explore the Agent API

03

Code Buckets

Encrypted storage for code and assets, organized by project. Up to 10 GB per bucket.

Explore Code Buckets

Writing & updates

Read along as we work.

Careers

Join the lab.

Engineers, researchers, DevOps, designers and writers, fully remote. If nothing fits, write to us anyway.

From an idea to a working AI system.