Tiny Recursive Model (TRM-Att)
No estimate
No hardware requirements for this model
The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.
On record
Full specification
Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.
Origin
Who built this model, where, and when it was published.
- Organisation
- Samsung SAIT AI Lab
- Organisation type
- Industry
- Country
- Korea (Republic of)
- Published
- 6 October 2025
- Authors
- Alexia Jolicoeur-Martineau
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language, Vision, Multimodal
- Task
- Language modeling/generation, Question answering, Visual puzzles
Size
How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.
- Parameters
- 7M
- Training data
- tokens
7M
Training compute
The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.
- Training compute
- 3.1 × 10²⁰ FLOP
- How it was established
- Hardware
989400000000000 FLOP/GPU/sec * 72 hours * 3600 sec / hour * 4 GPUs * 0.3 [assumed utilization] = 3.07742976e20 FLOP
The training run
What it physically took to train: which chips, how many, for how long, and what that drew from the wall.
- Training hardware
- NVIDIA H100 SXM5 80GB
- Chips used
- 4
- Wall-clock time
- 72 hours
- Power draw
- 5.5 kW
"Experiments on ARC-AGI were ran for around 3 days with 4 H100 with 80Gb of RAM."
Availability
Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.
- Weights
- Closed — provider access only
- Model access
- Unreleased
- Training code
- Open source
MIT license https://github.com/SamsungSAILMontreal/TinyRecursiveModels
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Confident
Sources
Where this record came from and when it was last checked.
- Reference
- Less is More: Recursive Reasoning with Tiny Networks
- Last updated
- 11 February 2026
What the numbers mean
Background
Tiny Recursive Model (TRM-Att) was published by Samsung SAIT AI Lab, in the country recorded as Korea (Republic of), during October 2025. The publishing organisation is categorised as industry.
It works in the domain of Language, Vision, Multimodal, and is recorded as performing the task of language modeling/generation, Question answering, Visual puzzles.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
How it was trained
The training run consumed about 3.1 × 10²⁰ FLOP, on hardware recorded as NVIDIA H100 SXM5 80GB. That figure measures what producing the model cost, and has no bearing on how fast it answers.
Answers
Tiny Recursive Model (TRM-Att) — common questions
Tiny Recursive Model (TRM-Att)— who created it?
It was published by Samsung SAIT AI Lab, based in Korea (Republic of), an organisation categorised as industry.
Tiny Recursive Model (TRM-Att)— when was it released?
It was published in October 2025.
Tiny Recursive Model (TRM-Att)— what is it used for?
It works in the domain of Language, Vision, Multimodal, and is recorded as handling the task of language modeling/generation, Question answering, Visual puzzles. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.
Tiny Recursive Model (TRM-Att)— how much compute was used to train it?
Training consumed around 3.1 × 10²⁰ FLOP, on hardware recorded as NVIDIA H100 SXM5 80GB. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.
Tiny Recursive Model (TRM-Att)— what GPU do I need to run it?
None. This is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.
Tiny Recursive Model (TRM-Att)— is it open source?
No. Its weights have not been published, so it exists only as a service controlled by its owner.
Tiny Recursive Model (TRM-Att)— how many parameters does it have?
It has a parameter count of 7M. 7M. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.
The other direction
Looking at it from the other side?
This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.