Research
Recurrent Looped Transformers: Pushing the Frontier in Decision Models and Scaling Reasoning in Latent Space
How recurrent looped transformers change GLiDE decision models, with experiments on decision quality, model size, and the cost of extra computation.
Research
The GLiDE Family: Decision Models from 0.4B to 27B
Meet five open weight GLiDE decision models from 0.4B to 27B. See benchmark results and how to choose a model for your application.
Research
Open-Weight Models in Production: Lessons From Our Panel
Lessons from Fastino Labs’ panel on choosing models, using post-training for customization, and putting small open-weight models to work.
Guide
Building Agents That Remember: A Guide to Agent Memory Solutions
Compare agent memory solutions for storing, updating, and retrieving context. Learn what to evaluate before adding memory to your AI application.
Research
How to Train a SOTA Frontier Decision Model
Discover how Fastino trained GLiDE v1.2, a state-of-the-art decision model, using reasoning distillation, improved training data, and multi-teacher learning to boost accuracy and reduce latency.
Guide
What is RLCD? The Training Method Behind Decision Models
Learn what reinforcement learning for calibrated decisions is, how it works, and when you should use it for your models.
Research
Big Model, Small Adapter: Fine-Tuning NVIDIA Nemotron 3.5 Super VL
See how LoRA fine-tuning NVIDIA Nemotron 3.5 Super VL improves finance and healthcare benchmarks while reducing output tokens and inference costs.
Product
Introducing the Fastino Training API: An Open Weight Models Training API for Supervised Fine-Tuning and Reinforcement Learning
Run supervised fine-tuning and reinforcement learning with the Fastino Training API, featuring managed GPUs, per-token pricing, and Tinker SDK compatibility.
Research
A Practical Guide to Harness Engineering
Learn harness engineering for AI agents: design tools, manage context and memory, enforce constraints, and benchmark accuracy, latency, and cost.
Read our published research, written by our technical team, from new model releases to product research.
Read the papers
Subscribe to our newsletter
Be the first to hear about what's new at Fastino Labs, including new model releases, product launches, and upcoming events.