DeepSpeech2 (English)

Closed weights Baidu Research - Silicon Valley AI Lab 38M parameters December 2015

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
Baidu Research - Silicon Valley AI Lab
Organisation type
Industry
Country
United States of America
Published
8 December 2015
Authors
Dario Amodei, Rishita Anubhai, Eric Battenberg, Carl Case, Jared Casper, Bryan Catanzaro, Jingdong Chen, Mike Chrzanowski, Adam Coates, Greg Diamos, Erich Elsen, Jesse Engel, Linxi Fan, Christopher Fougner, Tony Han, Awni Hannun, Billy Jun, Patrick LeGresley, Libby Lin, Sharan Narang, Andrew Ng, Sherjil Ozair, Ryan Prenger, Jonathan Raiman, Sanjeev Satheesh, David Seetapun, Shubho Sengupta, Yi Wan…

What it does

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

Domain
Speech
Task
Speech recognition (ASR)
Numerical format
FP32

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
38M

All networks have 38 million parameters.

Training data
716,400,000 tokens

"Our English speech system is trained on 11,940 hours of speech, while the Mandarin system is trained on 9,400 hours." 11,940 * 13,680 = 163339200

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
2.6 × 10¹⁹ FLOP

1 timestep = (1280 hidden units)^2 * (7 RNN layers * 4 matrices for bidirectional + 2 DNN layers) * (2 for doubling parameters from 36M to 72M) = 98 MFLOP 20 epochs * 12,000 hours * 3600 seconds/hour * 50 samples/sec * 98 MFLOP * 3 add-multiply * 2 backprop = 26,000 PF = 0.30 pfs-days See also AI and Compute by Dario Amodei and OpenAI https://openai.com/research/ai-and-compute

How it was established
Operation counting,Third-party estimation

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 GeForce GTX TITAN X
Chips used
16
Chip-hours
1,920
Wall-clock time
120 hours

"5 days" from AI and Compute https://openai.com/index/ai-and-compute/

Hardware utilisation
HFU 46.7%

"Overall the system sustains approximately 50 teraFLOP/second when training on 16 GPUs. This amounts to 3 teraFLOP/second per GPU which is about 50% of peak theoretical performance" HFU = (50/16) TFLOP / 6.69 TFLOP = 0.4670

Power draw
8.5 kW
Compute cost
$214

How it is classified

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

Frontier model
Yes
Record confidence
Confident
Citations
3,150

Sources

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

Reference
Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
Last updated
25 May 2026

What the numbers mean

About this model

DeepSpeech2 (English) was published by Baidu Research - Silicon Valley AI Lab, in the country recorded as United States of America, during December 2015. The category the publisher falls under is industry.

It works in the domain of Speech, and is recorded as performing the task of speech recognition (ASR).

This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.

What went into building it

Training it took a computation budget of roughly 2.6 × 10¹⁹ FLOP, on hardware recorded as NVIDIA GeForce GTX TITAN X. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Training consumed a corpus of around 716,400,000 tokens of text.

Answers

DeepSpeech2 (English) — common questions

01

DeepSpeech2 (English)— who created it?

It was published by Baidu Research - Silicon Valley AI Lab, based in United States of America, an organisation categorised as industry.

02

DeepSpeech2 (English)— when was it released?

It was published in December 2015. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

03

DeepSpeech2 (English)— what is it used for?

It works in the domain of Speech, and is recorded as handling the task of speech recognition (ASR). 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.

04

DeepSpeech2 (English)— how much compute was used to train it?

Training consumed around 2.6 × 10¹⁹ FLOP, on hardware recorded as NVIDIA GeForce GTX TITAN X. 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.

05

DeepSpeech2 (English)— 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.

06

DeepSpeech2 (English)— 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.

07

DeepSpeech2 (English)— how many parameters does it have?

It has a parameter count of 38M. All networks have 38 million parameters. 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.

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.