whatmodel.sh / Jurassic-1 Large (J1-Large, 7.5B)

Jurassic-1 Large (J1-Large, 7.5B)

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Jurassic-1 Large (J1-Large, 7.5B) is a closed-weights language model from AI21 Labs, released August 2021. Its published size is 7.5B parameters. Licence: proprietary.

Catalogue identity notes

AI21 Labs' J1-Large (7.5B per the white paper's Table 1; the abstract rounds to 7B). API-only, weights never released, later retired. Comparison-only.

Model information sources

Release date ↗

35.0±10.7comparison only: no current run option verifiedfrom 7 benchmark results across 1 source

Score within the ARC Challenge comparison set only.

$ how it compares

Scored LLMs released closest to Jurassic-1 Large (J1-Large, 7.5B), out of 1182 with a score, then more from AI21 Labs.

#modelscorewhereRAMreleased
979T0+ (T0p) 11B BigScience38±9comparison only2021-10
1027Codex code-cushman-001 OpenAI35±5comparison only2021-08
1028T0 11B BigScience35±3comparison only2021-10
1050Codex code-davinci-001 OpenAI33±1comparison only2021-08
1121GPT-J 6B EleutherAI25±1local~14GB2021-06
more from AI21 Labs
325Jamba 1.6 Large AI21 Labs59±11local~839GB2025-02
495Jamba 1.6 Mini AI21 Labs54±10local~34GB2025-02
546Jamba 1.5 Large AI21 Labs52±1local~839GB2024-08

see all LLMs · find the ones that fit your machine

$ what it's good at

other tests8 not in a task · 38% of its score
testresultvs the others in its rundifficultycountswhy
PIQA lab78.8% ✎▁▁▁▂▂▃▅█▅▃▁▁ beats 83/123medium · avg 80%▇▇▇▇ 9.4%sharp · at its level · small run
Story Cloze Test lab +180.2% ✎▃▃▃▂▅█▅▃▁··▂ beats 43/50medium · avg 82%▇▇▇ 8.4%sharp · at its level · small run
ARC Easy lab62.2% ✎▁▃▂▄▆█▅▃▁··▁ beats 37/100medium · avg 65%▇▇▇ 8.1%at its level · small run
WinoGrande lab64.4% ✎▁▁▁▁▂▁▁▅▇█▃▁ beats 54/131hard · avg 67%▇▇ 4.5%at its level · small run
ARC Challenge lab41.7% ✎▄▁▁▂▁▄█▇▃▂▃▂ beats 58/140hard · avg 45%▇▇ 3.9%small run
HellaSwag lab71.9% ✎▁▁▁▂▂▂▄▆█▇▄▂ beats 76/135easy · avg 75%▇ 3.7%at its level · small run
BoolQ lab65.0% ✎▁····▁▂█▆▅▆▄ beats 36/103medium · avg 70%▇ 0.4%blunt · at its level · small run
OpenBookQA lab46.0% ✎▅▁▁▁▂▂▃█▃▁▃▁ beats 53/107—not counted yet: no other model in the same run to compare it with

$ where to run it

No verified API or download recorded yet · add one

$ how the score was worked out · AI21 Labs official docs 38%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
AI21 Labs official docs · lab-reported7Factual first-party specification (not licensed content)
prior · pull toward the middle
weighted average= 35.0

not counted (2)

Parallel Context Windows for Large Language Models — authors' own runsOpenBookQA [correct]46.0%fewer than three models have this test yet, or this source ran it for this model alone
Parallel Context Windows for Large Language Models — authors' own runsStory Cloze Test [correct]84.2%the only model in this source's run of the test, so there is nothing to compare it with there; another run of the test counts

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json