CODEFUSION (Python)

Closed weights Microsoft,Microsoft Research 75M parameters October 2023

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
Microsoft,Microsoft Research
Organisation type
Industry,Industry
Country
United States of America
Published
26 October 2023
Authors
Mukul Singh, José Cambronero, Sumit Gulwani, Vu Le, Carina Negreanu, Gust Verbruggen

What it does

The problem areas the model was built for. A model can carry several of each.

Domain
Language
Task
Code generation
Approach
Self-supervised learning

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
75M

Table 1

Training data
4,390,400 tokens

Section A3, Table 5: for python, 56k samples with an average length of 78.4 tokens

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
7.9 × 10¹⁸ FLOP

V100 performance: 125 teraFLOPS according to https://www.nvidia.com/en-us/data-center/v100/ 11 hours * 4 GPUs * 125 teraFLOPS/GPU * 0.40 utilization = 7.92e18 FLOP

How it was established
Hardware

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 V100
Chips used
4
Wall-clock time
11 hours

"The system used to run the experiments uses an Intel Core i7 processor (base at 1.8 GHz) along with 4 V100 GPU units, a 64-bit operating system, and 56 GB RAM. CODEFUSION took 8 hours to pre-train and 3 hours to fine-tune on average for each dataset."

Power draw
2.4 kW
Compute cost
$9

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
Unreleased

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Why it is tracked
SOTA improvement

See Table 1, SOTA in Python code generation "We evaluate Python using CodeBERTScore (Zhou et al., 2023), which has been shown to be a high quality non-execution-based code matching metric."

Record confidence
Confident
Citations
57

Sources

Where this record came from and when it was last checked.

Reference
CODEFUSION: A Pre-trained Diffusion Model for Code Generation
Last updated
25 May 2026

What the numbers mean

About this model

CODEFUSION (Python) was published by Microsoft,Microsoft Research, in United States of America, in October 2023. The organisation is categorised as industry,Industry.

It works in Language, and is recorded as doing code generation.

Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.

How it was trained

Training it took roughly 7.9 × 10¹⁸ FLOP of computation, on NVIDIA V100 — a measure of what producing the model cost, not of how fast it answers.

The training set ran to roughly 4,390,400 tokens.

It is tracked in the underlying dataset for one reason in particular: sOTA improvement.

Answers

CODEFUSION (Python) — common questions

01

How much compute was used to train CODEFUSION (Python)?

Around 7.9 × 10¹⁸ FLOP, on NVIDIA V100. 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.

02

What GPU do I need to run CODEFUSION (Python)?

None. CODEFUSION (Python) 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.

03

Is CODEFUSION (Python) open source?

No. CODEFUSION (Python) has not had its weights published, so it exists only as a service controlled by its owner.

04

How many parameters does CODEFUSION (Python) have?

CODEFUSION (Python) has 75M parameters. Table 1. 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.

05

Who created CODEFUSION (Python)?

CODEFUSION (Python) was published by Microsoft,Microsoft Research, based in United States of America, categorised as industry,Industry.

06

When was CODEFUSION (Python) released?

CODEFUSION (Python) was published in October 2023. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

07

What is CODEFUSION (Python) used for?

CODEFUSION (Python) works in Language, and is recorded as handling code generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Source

Original publication

Record last updated 25 May 2026

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.