SOTA sub-1B coding models distilled from execution-verified traces. Optimized for local code generation, debugging, and agentic software engineering o
AI & ML interests
Building open-source AI infrastructure, SOTA edge models, and evaluation tooling. Intelligence isn't about scale; it's about precision | LLM Fine-Tuning | Knowledge Distillation | MLOps & CI/CD Tooling | Vision-Language Models (VLMs) | Reasoning Models |
Recent Activity
Organization Card
🪐 Pluto AI Labs
Building open-source AI infrastructure, frontier edge models, and evaluation tooling.
🚀 Our Mission
The AI industry is obsessed with scaling up. We are obsessed with scaling down.
Pluto AI Labs is an open-source AI research lab focused on proving that frontier-level reasoning, coding, and vision capabilities can be distilled into compact models that run on consumer hardware.
We build the models, datasets, evaluation systems, and MLOps infrastructure required to bring capable AI to the edge — running locally on laptops, phones, and offline servers without compromising performance, quality, or privacy.
🛠️ The Ecosystem
We don't just train models. We build the infrastructure to evaluate, distill, and deploy them.
| Collection | Description | Status |
|---|---|---|
| 🧬 Apollo-Reasoning | Frontier reasoning traces distilled into compact models for local reasoning, mathematics, and complex task solving. | 🔬 Active Research |
| 🛠️ Atlas-Coding | High-performance sub-1B coding models optimized for efficient local inference and evaluated on strict coding benchmarks. | 🚀 Active |
| 🧠 Pluto-General | General-purpose edge LLMs designed for multi-step reasoning, instruction following, and task completion. | ✅ Released |
| 👁️ Apollo-VL | Compact vision-language models focused on visual reasoning, OCR, charts, documents, and UI understanding. | 🔬 In Development |
| 🧰 Tools & Datasets | llm-diff behavioral evaluation tooling and high-signal datasets for model distillation and training. |
🚀 Active |
All while remaining practical for offline inference on consumer hardware.
📦 Research Collections
Our work spans several layers of the efficient-AI stack:
Models → Datasets → Distillation → Evaluation → MLOps → Edge Deployment
We release research artifacts openly whenever possible so others can reproduce, evaluate, and build upon our work.
📄 Research & Whitepapers
We believe useful AI research should be measurable, reproducible, and open.
📘 The Distillation Bottleneck
An empirical study of behavioral changes across successive LLM releases and the limitations of evaluating model improvements through benchmarks alone.
📘 Pluto-Genesis
Research on reliable QLoRA fine-tuning under ephemeral compute constraints, including cross-session checkpoint recovery for interrupted training workflows.
🤝 Connect & Contribute
Pluto AI Labs is an independent, open-source research lab built around a simple idea:
Capability should not be limited by access to massive compute.
We welcome researchers, engineers, contributors, and builders interested in efficient AI, model distillation, multimodal reasoning, edge inference, and open-source machine learning.
🔗 Links
- 🌐 GitHub: https://github.com/Pluto-AI-Labs
- 🤗 Hugging Face: https://huggingface.co/Pluto-AI-Labs
- 👤 Founder: Siddharth N.R.
“Intelligence isn't about scale. It's about precision.”
Pluto AI Labs — Open Models · Open Research · Efficient Intelligence
SOTA sub-1B coding models distilled from execution-verified traces. Optimized for local code generation, debugging, and agentic software engineering o
General-purpose edge LLMs fine-tuned for multi-step reasoning, mathematics, and conversational task completion. Optimized to run efficiently on consum
models 7
Pluto-AI-Labs/Atlas-Frontier-Distill-3B
Text Generation • 3B • Updated • 47
Pluto-AI-Labs/Atlas-Coder-2-0.5B
Text Generation • 0.5B • Updated • 254 • 1
Pluto-AI-Labs/Atlas-Coder-2-0.5B-GGUF
Text Generation • 0.5B • Updated • 282
Pluto-AI-Labs/Atlas-Coder-0.5B
Text Generation • 0.5B • Updated • 75 • 1
Pluto-AI-Labs/Atlas-Coder-0.5B-GGUF
Text Generation • 0.5B • Updated • 172
Pluto-AI-Labs/Pluto-Genesis-0.6B-GGUF
Text Generation • 0.6B • Updated • 443
Pluto-AI-Labs/Pluto-Genesis-0.6B
Text Generation • 0.6B • Updated • 1.68k