Codesota · Models · Gemini UltraGoogle DeepMind3 results · 3 benchmarks
Model card

Gemini Ultra.

Google DeepMindproprietaryUnknown paramsTransformer (decoder-only)

Largest Gemini 1.0 model. Released December 2023.

§ 01 · Benchmarks

Every benchmark Gemini Ultra has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01SNLINatural Language Processing · Natural Language Inferenceaccuracy91.9%#3/82023-12-19source ↗
02SuperGLUENatural Language Processing · Text Classificationaverage-score90.0%#4/72023-12-19source ↗
03GSM8KReasoning · Mathematical Reasoningaccuracy94.4%#22/322024-02-01source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 02 · Strengths by area

Where Gemini Ultra actually performs.

Natural Language Processing
2
benchmarks
avg rank #3.5
Reasoning
1
benchmark
avg rank #22.0
§ 03 · Papers

1 paper with results for Gemini Ultra.

  1. 2023-12-19· Natural Language Processing· 2 results

    Gemini: A Family of Highly Capable Multimodal Models

§ 04 · Related models

Other Google DeepMind models scored on Codesota.

Gemma 3 12B IT
12B params · 2 results
Gemma 3 4B IT
4B params · 2 results
Disco57
1 result
DreamerV3
1 result
MEME
1 result
SoViT-400m/14
400M params · 1 result
BBF (Bigger, Better, Faster)
Unknown params · 0 results
§ 05 · Sources & freshness

Where these numbers come from.

arxiv
2
results
gsm8k-shadow-page
1
result
2 of 3 rows marked verified. · first result 2023-12-19, latest 2024-02-01.