TD-Gammon
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
- IBM
- Organisation type
- Industry
- Country
- United States of America
- Published
- 1 May 1992
- Authors
- G Tesauro
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Games
- Task
- Backgammon
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
- 25K
- Training data
- 6,300,000 tokens
"The best performance was obtained with a network containing 80 hidden units and over 25,000 weights."
"This network was trained for over 300,000 training games" Each backgammon game has an avg of around 21 movements https://www.bkgm.com/rgb/rgb.cgi?view+712
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
- 1.8 × 10¹³ FLOP
- How it was established
- Third-party estimation,Operation counting,Hardware
Extracted from AI and Compute (https://openai.com/blog/ai-and-compute/) charts by using https://automeris.io/WebPlotDigitizer/. OpenAI estimate: 1.8e13 Hardware estimate (likely overestimates due to simulation effort) "on an IBM RS/6000 workstation, the smallest network was trained in several hours, while the largest net required two weeks of simulation time." IBM RS/6000 achieves 1.5 GFLOPS on Linpack (https://link.springer.com/rwe/10.1007/978-0-387-09766-4_232) 14*24*60*60*0.5*1500000000=9.…
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Frontier model
- Yes
- Why it is tracked
- Highly cited,Historical significance
- Record confidence
- Speculative
- Citations
- 1,344
Sources
Where this record came from and when it was last checked.
- Reference
- Practical Issues in Temporal Difference Learning
- Last updated
- 28 November 2025
What the numbers mean
Background
TD-Gammon was published by IBM, in the country recorded as United States of America, during May 1992. The category the publisher falls under is industry.
It works in the domain of Games, and is recorded as performing the task of backgammon.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
Training and provenance
Producing it required arithmetic totalling around 1.8 × 10¹³ FLOP. That figure measures what producing the model cost, and has no bearing on how fast it answers.
Training consumed a corpus of around 6,300,000 tokens of text.
The reason it appears in this catalogue at all: highly cited,Historical significance.
Answers
TD-Gammon — common questions
TD-Gammon— what is it used for?
It works in the domain of Games, and is recorded as handling the task of backgammon. 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.
TD-Gammon— how much compute was used to train it?
Training consumed around 1.8 × 10¹³ FLOP. 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.
TD-Gammon— 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.
TD-Gammon— is it open source?
The licensing was never recorded in our source data. We treat unstated licensing as closed, because an unrecorded licence is not one to rely on.
TD-Gammon— how many parameters does it have?
It has a parameter count of 25K. "The best performance was obtained with a network containing 80 hidden units and over 25,000 weights.". 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.
TD-Gammon— who created it?
It was published by IBM, based in United States of America, an organisation categorised as industry.
TD-Gammon— when was it released?
It was published in May 1992. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.
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.