Training/fine-tuning stack — frameworks, compute, RLHF tooling, model post-training.
16 verified facts across 12 companies
Modal Executes reinforcement learning rollouts, grading, and inference workloads.
Receipts · 1
“Applied Compute makes use of Modal Functions to provide inexpensive serverless fan-out without requiring a dedicated cluster.” — modal.com, May 20, 2026
Source last verified Jul 27, 2026
Together GPU Clusters
by Together AI Trains voice models on multi-node clusters scheduled via Slurm.
Receipts · 1
“Cartesia also uses Together’s H100 80GB clusters for model training, with multi-node workloads scheduled via Slurm.” — together.ai, Jul 2026
Source last verified Jul 27, 2026
Ray
by Anyscale (open source) Underpins the custom asynchronous reinforcement-learning training infrastructure with PyTorch.
Receipts · 2
“We built custom training infrastructure leveraging PyTorch and Ray to power asynchronous reinforcement learning at scale.” — cursor.com, Oct 29, 2025
“Build our distributed training, inference, and RL infrastructure” — cursor.com, Jul 2026
Source last verified Jul 26, 2026
NVIDIA GPUs (Blackwell)
by NVIDIA Trains mixture-of-experts models at low precision on large GPU clusters.
Receipts · 2
“allowing us to scale training to thousands of NVIDIA GPUs with minimal communication cost” — cursor.com, Oct 29, 2025
“Custom low-precision kernels for efficient MoE training on Blackwell GPUs” — cursor.com, Mar 27, 2026
Source last verified Jul 26, 2026
Internal compute platform running sandboxed cloud environments for RL rollouts.
Receipts · 2
“Anyrun, our internal compute platform for running hundreds of thousands of sandboxed coding environments” — cursor.com, Mar 27, 2026
“running hundreds of thousands of concurrent sandboxed coding environments in the cloud” — cursor.com, Oct 29, 2025
Source last verified Jul 26, 2026
Trains the Tab model continuously on live accept and reject feedback.
Receipts · 2
“rolling out new models to users frequently throughout the day and using that data for training” — cursor.com, Sep 12, 2025
“train frontier coding agents and scale RL on real user data” — cursor.com, Jul 2026
Source last verified Jul 26, 2026
Together GPU Clusters
by Together AI Trains large open-weight models on multi-node GPU clusters.
Receipts · 1
“Deep Cogito uses Together for customizable H100/H200 GPU clusters, reliable long-run training infrastructure” — together.ai, Jul 2026
Source last verified Jul 27, 2026
Fireworks AI Trains large open models via reinforcement fine-tuning for Deep Research.
Receipts · 1
“By leveraging Fireworks’ Reinforcement Fine Tuning to train large state-of-the-art open models” — fireworks.ai, Oct 31, 2025
Source last verified Jul 27, 2026
Together GPU Clusters
by Together AI Trains clinical models with multi-node RL and long-context runs.
Receipts · 1
“Latent Health chose Together Instant Clusters for clinical AI training” — together.ai, Jul 2026
Source last verified Jul 27, 2026
Fine-tunes small open-weight models for search, routing, and function calling.
Receipts · 2
“By fine-tuning models, we reduced latency from about 2 seconds to 350 milliseconds” — fireworks.ai, Jul 25, 2025
“Our, um, fine tuned and open source models are served on GPUs, right?” — latent.space, Apr 15, 2026
Source last verified Jul 26, 2026
Amazon SageMaker HyperPod
by AWS Early distributed training platform, since replaced by a self-managed GPU fleet.
Receipts · 2
“Amazon SageMaker HyperPod's built-in data and model parallel libraries helped us optimize training time on GPUs” — aws.amazon.com, Mar 2024
“Perplexity Accelerates Foundation Model Training by 40% with Amazon SageMaker HyperPod” — youtube.com, Mar 27, 2024
Source last verified Jul 26, 2026
NeMo
by NVIDIA Framework behind the post-training run that produced R1-1776.
Receipts · 1
“They used NVIDIA's NeMo 2.0 framework to fine-tune the model” — hyperight.com, Feb 24, 2025
Source last verified Jul 26, 2026
Together GPU Clusters
by Together AI Provides bare-metal GPU infrastructure for custom model training.
Receipts · 1
“How Scaled Cognition Trains APT-1 on Together AI GPU Clusters” — together.ai, Jul 2026
Source last verified Jul 27, 2026
Post-trains Sidekick's fine-tuned models with LLM judges as reward signals.
Receipts · 2
“we implemented Group Relative Policy Optimization (GRPO), a reinforcement learning approach that uses our LLM judges as reward signals” — shopify.engineering, Aug 26, 2025
“Fine-tuning an open-source model got us there for the common cases.” — shopify.engineering, Jun 15, 2026
Source last verified Jul 26, 2026
Together AI Runs supervised fine-tuning and preference optimization pipelines.
Receipts · 1
“using Together AI for key components” — together.ai, Jul 2026
Source last verified Jul 27, 2026
Together Fine-Tuning Platform
by Together AI Trains customer-specific Qwen models for EOB parsing.
Receipts · 1
“XY.AI migrated from its self‑hosted stack to the Together Fine‑Tuning Platform” — together.ai, Jul 2026
Source last verified Jul 27, 2026