alt="drawing" width="600"/> If you already know T5, FLAN-T5 is just better at everything. For the same number of parameters, these models have been fine-tuned on more than 1000 additional tasks covering also more languages.
Runs On
What it takes to serve flan-t5-large (783M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 1.6 GB | 1.9 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.8 GB | 0.9 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.4 GB | 0.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
SAVRN's Notes on flan-t5-large
Twelve files and 12.8 GB read like a far bigger model than 783M parameters. The repository carries safetensors, pytorch, jax and tf copies, and a deployment loads one: at 16-bit that is 1.6 GB of weights and 1.9 GB of memory needed. The cheapest Index setup, one MI300X with 192 GB at $1.85 per hour on-demand, is far more card than this model can use alone. The dataset relations name gsm8k, aqua_rat, esnli, deepmind/code_contests, lambada, taskmaster2, svakulenk0/qrecc and djaym7/wiki_dialog, a tell for where the 1000-plus fine-tuning tasks lean.
Apache 2.0 covers commercial use, modification and redistribution as long as the notices travel with it and significant changes are stated. Before committing, check three gaps in the record: no context length is listed, no task tag is set, and the last update was 2023-07-17. The Index shows no per-token host prices, so your own run is the price.
Model Card
By Google, published under apache-2.0, revision 0613663d0d48.
Model Card for FLAN-T5 large
Table of Contents
- TL;DR
- Model Details
- Usage
- Uses
- Bias, Risks, and Limitations
- Training Details
- Evaluation
- Environmental Impact
- Citation
- Model Card Authors
TL;DR
If you already know T5, FLAN-T5 is just better at everything. For the same number of parameters, these models have been fine-tuned on more than 1000 additional tasks covering also more languages. As mentioned in the first few lines of the abstract :
Flan-PaLM 540B achieves state-of-the-art performance on several benchmarks, such as 75.2% on five-shot MMLU. We also publicly release Flan-T5 checkpoints,1 which achieve strong few-shot performance even compared to much larger models, such as PaLM 62B. Overall, instruction finetuning is a general method for improving the performance and usability of pretrained language models.
Disclaimer: Content from this model card has been written by the Hugging Face team, and parts of it were copy pasted from the T5 model card.
Model Details
Model Description
Configuration
- Architecture
- T5ForConditionalGeneration
- Vocabulary size
- 32,128
- Model type
- t5
Identity and Version
- Repository
- google/flan-t5-large
- Publisher
- Task
- Not stated by the source
- Modality
- Other
- Library
- transformers
- Parameters
- 783M parameters
- Languages
- en, fr, ro, de
- Revision
- 0613663d0d48ea86ba8cb3d7a44f0f65dc596a2a
- First published
- 2022-10-21
- Last updated
- 2023-07-17
Files and Weights
12 files, 12.8 GB in total. The weights are 4 files totalling 12.8 GB in bin, h5, msgpack, safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| flax_model.msgpack | Weights | 3.1 GB | 44c5f419a8cb |
| model.safetensors | Weights | 3.1 GB | a9dd06ce490f |
| pytorch_model.bin | Weights | 3.1 GB | 93249e5f2b05 |
| tf_model.h5 | Weights | 3.4 GB | 1e41a6feb46b |
| config.json | Configuration | 662 B | — |
| generation_config.json | Configuration | 147 B | — |
| special_tokens_map.json | Configuration | 2.2 KB | — |
| README.md | Documentation | 10.8 KB | — |
| .gitattributes | Repository | 1.4 KB | — |
| spiece.model | Tokenizer | 791.7 KB | d60acb128cf7 |
| tokenizer.json | Tokenizer | 2.4 MB | — |
| tokenizer_config.json | Tokenizer | 2.5 KB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 12.8 GB
Released by Google through Kaggle. Read the license.
Built From
- Described by arXiv:1910.09700
- Described by arXiv:2210.11416
- Trained on (disclosed) aqua_rat
- Trained on (disclosed) deepmind/code_contests
- Trained on (disclosed) djaym7/wiki_dialog
- Trained on (disclosed) esnli
- Trained on (disclosed) gsm8k
- Trained on (disclosed) lambada
- Trained on (disclosed) qed
- Trained on (disclosed) quasc
- Trained on (disclosed) svakulenk0/qrecc
- Trained on (disclosed) taskmaster2
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 12.8 GB |
| 16-bit | 1.6 GB |
| 8-bit | 0.8 GB |
| 4-bit | 0.4 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About flan-t5-large
How much GPU memory does flan-t5-large need?
About 1.9 GB at 16-bit and 0.5 GB at 4-bit: the weights (783M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run flan-t5-large on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use flan-t5-large commercially?
Yes. flan-t5-large is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.