Reka Flash

Closed weights Reka AI 21B parameters April 2024

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
Reka AI
Organisation type
Industry
Country
United States of America
Published
15 April 2024
Authors
Aitor Ormazabal, Che Zheng, Cyprien de Masson d'Autume, Dani Yogatama, Deyu Fu, Donovan Ong, Eric Chen, Eugenie Lamprecht, Hai Pham, Isaac Ong, Kaloyan Aleksiev, Lei Li, Matthew Henderson, Max Bain, Mikel Artetxe, Nishant Relan, Piotr Padlewski, Qi Liu, Ren Chen, Samuel Phua, Yazheng Yang, Yi Tay, Yuqi Wang, Zhongkai Zhu, Zhihui Xie

What it does

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

Domain
Multimodal, Language, Vision, Video, Speech
Task
Chat, Language modeling/generation, Image captioning, Code generation, Code autocompletion, Question answering, Visual question answering, Video description, Speech recognition (ASR), Speech-to-text

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
21B

21B dense

Training data
tokens

5T text tokens + unknown amount of multimodal data

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
6.3 × 10²³ FLOP

Reka Flash has 21B parameters and was trained on 5 trillion language tokens (+ unknown amount of multimodal data) 6 FLOP / token / parameter * 21B parameters * 5 trillion text tokens = 6.3 × 10^23 FLOP OOM agrees with GPU details: "Reka Flash and Edge were trained on several hundreds of H100s across a period of several weeks." 3 weeks * 300 H100s * 7 day/week * 24 hour/day * 3600 s/day * 9.9e14 FLOP/GPU-s * 0.3 [assumed utilization]= 1.6156506e+23 FLOP Not enough info to estimate SFT and RL…

How it was established
Hardware,Operation counting

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 A100,NVIDIA H100 SXM5 80GB

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
API access
Training code
Unreleased

How it is classified

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

Likely above 10²³ FLOP
Yes
Record confidence
Likely

Sources

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

Reference
Reka Core, Flash, and Edge: A Series of Powerful Multimodal Language Models
Last updated
28 November 2025

What the numbers mean

What this model is

Reka Flash was published by Reka AI, in United States of America, in April 2024. The organisation is categorised as industry.

It works in Multimodal, Language, Vision, Video, Speech, and is recorded as doing chat, Language modeling/generation, Image captioning, Code generation, Code autocompletion, Question answering, Visual question answering, Video description, Speech recognition (ASR), Speech-to-text.

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

Training and provenance

Producing it required around 6.3 × 10²³ FLOP of arithmetic, on NVIDIA A100,NVIDIA H100 SXM5 80GB, which is a statement about the training budget rather than about inference.

Answers

Reka Flash — common questions

01

What is Reka Flash used for?

Reka Flash works in Multimodal, Language, Vision, Video, Speech, and is recorded as handling chat, Language modeling/generation, Image captioning, Code generation, Code autocompletion, Question answering, Visual question answering, Video description, Speech recognition (ASR), Speech-to-text. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

02

How much compute was used to train Reka Flash?

Around 6.3 × 10²³ FLOP, on NVIDIA A100,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.

03

What GPU do I need to run Reka Flash?

None. Reka Flash 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.

04

Is Reka Flash open source?

No. Reka Flash has not had its weights published, so it exists only as a service controlled by its owner.

05

How many parameters does Reka Flash have?

Reka Flash has 21B parameters. 21B dense. 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.

06

Who created Reka Flash?

Reka Flash was published by Reka AI, based in United States of America, categorised as industry.

07

When was Reka Flash released?

Reka Flash was published in April 2024. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

Source

Original publication

Record last updated 28 November 2025

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.