opt-6.7b

Text in, text out, 2K tokens of context.

Verified 2026-10-25 Rank #524 in LLMs · score 75.7 method v1 Sources 3 Downloads 30d 47.8k · #385 of 619 tracked

Can I use it?OtherCustom license; read the repo LICENSE before commercial hostinglicence file2026-10-24

Do people use it?47.8kdownloads, 30 days#385 of 619 we track · below the median+1 likes, downloads +60% this week124 likesHugging Face2026-10-24

Where it's used

Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.

No API provider serves it, so its downloads are people running it themselves: 1 quantized builds and 37 Spaces.

As a service0no API provider listed
On own hardware1quantized builds
Built on8710 fine-tunes · 76 adapters · 1 merges
In apps37Hugging Face Spaces

Top Spaces running it: tomg-group-umd/lm-watermarking ♥ 100 · jykoh/fromage ♥ 54 · alvanlii/FROMAGe ♥ 27 · jykoh/gill ♥ 19

The story so far

Weights, every dated article, and repo commits on one rail.

21 repo commits since the weights landed

Teal is the weights (2022-05-11), grey ticks are repo commits.

What people built on it

88 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

88 derivatives on the Hub, most of them adapters

From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.

GPTQ iproskurina/opt-6.7b-GPTQ-4bit-g128 for vLLM, SGLang 1.2k downloads

All variants →

Reference

Everything else, folded: the full tables, every source.

What the card says

1 cited sentence

Summary generated, cited

OPT-6.7B is a decoder-only transformer model developed by facebook for text-generation tasks.^[1]^[2]^[3]

organization
pipeline tag
“decoder-only pre-trained transformers”

Generated 2026-10-07 15:13 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Facts

18 fields

Listed in LLMs

Facts

Context length is 2,048 tokens (2K). (config.json max_position_embeddings, observed 2026-10-24)
License: Other (friction: review). Custom license; read the repo LICENSE before commercial hosting. (license file, observed 2026-10-24)
Created 2022-05-11, last modified 2023-01-24. (HF API, observed 2026-10-24)

Pipeline tag	text-generation
Library	transformers
Architectures	OPTForCausalLM
Model type	opt
Dtype	float16
MoE	no
Context length	2,048 (config.json max_position_embeddings)
Languages	en
Papers	arXiv:2205.01068, arXiv:2005.14165
License	Other (other)
Hub snapshot	2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · OPTForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · OPTForCausalLM · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed

vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.

All variants

40 variants

Lineage

Original model Family: facebook/opt 88 derivatives on the Hub

quantized: 1finetune: 10adapter: 76merge: 1

Top variants (40 shown, ranked by downloads)

Variant	Relation	Quant	Params	Band	License	Author	Likes	Downloads
iproskurina/opt-6.7b-GPTQ-4bit-g128	quantized	GPTQ	1.29B	S	other	iproskurina	0	1.2k
MiniLLM/MiniLLM-OPT-6.7B	finetune	—	—	—	apache-2.0	MiniLLM	0	650
royokong/prompteol-opt-6.7b	adapter	—	—	—	—	royokong	0	514
mturck/opt-6.7b-lora-code-completion	adapter	—	—	—	—	mturck	7	107
pkarypis/opt-6.7b-sft	finetune	—	6.66B	M	other	pkarypis	0	384
jiangchengchengNLP/opt-6.7B-AWQ-8INT	finetune	AWQ	6.87B	M	apache-2.0	jiangchengchengNLP	0	225
longquan/opt-6.7b-lora-1102	adapter	—	—	—	—	longquan	1	133
Charleyhoo/opt-6.7b-lora	adapter	—	—	—	—	Charleyhoo	0	165
Uppaal/opt-ProFS-toxicity	finetune	—	6.86B	M	mit	Uppaal	0	150
caesar-one/opt-6.7b-lora-short-reviews	adapter	—	—	—	—	caesar-one	0	129
lisali126/opt-6.7b-lora	adapter	—	—	—	—	lisali126	0	121
farnooshazour/opt-6.7b-lora	adapter	—	—	—	—	farnooshazour	0	113
yuping322/opt-6.7b-lora	adapter	—	—	—	—	yuping322	0	112
TrustAI-lab/Auto-Redteaming-Model	adapter	—	—	—	—	TrustAI-lab	0	112
awesomenes/opt-6.7b-lora	adapter	—	—	—	—	awesomenes	0	107
dolo650/opt-6.7b-lora	adapter	—	—	—	—	dolo650	0	104
ramnathv/opt-6.7b-lora	adapter	—	—	—	—	ramnathv	0	99
Chow33/opt-6.7b-lora	adapter	—	—	—	—	Chow33	0	97
vhasghari/opt-6.7b-lora_Clinical	adapter	—	—	—	—	vhasghari	0	85
actionpace/opt-6.7b-lora	adapter	—	—	—	—	actionpace	0	83
kasiarun/opt-6.7b-lora	adapter	—	—	—	—	kasiarun	0	77
math8485/opt-6.7b-lora	adapter	—	—	—	—	math8485	0	77
sushant07/opt-6.7b-lora	adapter	—	—	—	—	sushant07	0	76
Praveendecode/facebook-opt-6.7b-lora-finetuned	adapter	—	—	—	—	Praveendecode	0	73
evolevelyn/opt-6.7b-lora	adapter	—	—	—	—	evolevelyn	0	70
xuanvuzzz260101/opt-6.7b-lora	adapter	—	—	—	—	xuanvuzzz260101	0	69
foobb/opt-6.7b-lora	adapter	—	—	—	—	foobb	0	67
sandeepksingh1/opt-6.7b-lora	adapter	—	—	—	—	sandeepksingh1	0	66
Jaynm31245/oo500	adapter	—	—	—	—	Jaynm31245	0	64
Numlock/sn-opt-peft	adapter	—	—	—	—	Numlock	0	63
EdBerg/opt-6.7b-lora	adapter	—	—	—	—	EdBerg	0	63
tad1234/opt-6.7b-lora	adapter	—	—	—	—	tad1234	0	62
joseph-a-trevorrow/opt-6.7b-lora	adapter	—	—	—	—	joseph-a-trevorrow	0	55
Puch/Databricks-lora	adapter	—	—	—	—	Puch	0	43
Jaynm31245/oo	adapter	—	—	—	—	Jaynm31245	0	40
Numlock/svn-opt-3	adapter	—	—	—	—	Numlock	0	39
Tatiana239/opt-6.7b-lora	adapter	—	—	—	—	Tatiana239	0	38
RMHalak/adapters-opt-bnb8-QLORA-super_glue-copa	adapter	-bnb	—	—	—	RMHalak	0	38
adenovirux/facebook-opt-6.7b-qlora	adapter	—	—	—	—	adenovirux	0	36
HamidBekam/opt-6.7b-lora	adapter	—	—	—	—	HamidBekam	0	35

Repository changelog

21 commits

Repository changelog 21 commits

Date	Commit	SHA
2023-01-24	Adding generation config file(s)	a45aa65
2022-06-24	Add Flax weights (#7)	8dc17cd
2022-06-24	Add TF weights (#5)	b58ac05
2022-06-22	Update config.json	e41a654
2022-06-22	Update README.md (#4)	40d1ea0
2022-06-21	correct checkpoints see: https://github.com/facebookresearch/metaseq/pull/164	d958377
2022-05-26	Change BOS token from 0 to 2 as BOS token is equal to EOS for OPT. See: https://github.com/huggingface/transformers/issues/17431 (#1)	b8089eb
2022-05-25	Update tokenizer_config.json	3a00ffb
2022-05-16	Set commercial attribute to False	b62fba2
2022-05-16	Update README	daaa626
2022-05-13	Temporarily deactivate the inference API	4074100
2022-05-13	Update config.json	b6d51b6
2022-05-12	Update README.md	1286852
2022-05-12	Update README.md	d1a95fd
2022-05-12	Update README.md	88cc4ae
2022-05-12	Update README.md	769c7b9
2022-05-12	Create README.md	a067f54
2022-05-12	finish	892f69f
2022-05-12	Update config.json	5eae555
2022-05-11	upload	2e2cb70
2022-05-11	initial commit	70e8164

Papers

2 papers

Papers 2

arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).

OPT: Open Pre-trained Transformer Language Models this model

Susan Zhang, Stephen Roller, Naman Goyal +16 · 2022-05-02 · PDF · HF paper page (3 upvotes)
Language Models are Few-Shot Learners cited in card peer-reviewed · measured

Tom B. Brown, Benjamin Mann, Nick Ryder +28 · 2020-05-28 · neural information processing systems · PDF · HF paper page (21 upvotes) · evidence

Similar models

3 models

Similar models

gpt-j-6b
same task (text-generation), permissive license
DialoGPT-medium
same task (text-generation), permissive license
Qwen-VL-Chat
same task (text-generation)

Sources

3 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:41 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:46 UTC · Summary: 2026-10-07 15:13 UTC · Last verified 2026-10-25

Follow opt-6.7b

Watch this model

/feeds/models/facebook--opt-6.7b.xml(RSS 2.0) · every change to opt-6.7b
/feeds/families/facebook--opt.xml(RSS 2.0) · every change across the opt family

Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.