Codesota · Models · CoCaGoogle1 results · 1 benchmarks
Model card

CoCa.

Googleopen-sourceUnknown paramsImage encoder + cross-attention + causal decoder

Contrastive Captioner. Trained on JFT-3B and LAION-2B. CIDEr 143.6 on COCO NoCaps. 2022. Source: arxiv:2205.01068.

§ 02 · Benchmarks

Every benchmark CoCa has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01COCO CaptionsMultimodal · Image CaptioningCIDEr143.60#2/22022-05-02source ↗
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.
§ 03 · Strengths by area

Where CoCa actually performs.

Multimodal
1
benchmark
avg rank #2.0
§ 04 · Papers

1 paper with results for CoCa.

  1. 2022-05-02· Multimodal· 1 result

    CoCa: Contrastive Captioners are Image-Text Foundation Models

§ 05 · Related models

Other Google models scored on Codesota.

Gemini 1.5 Pro
21 results
Gemini 2.5 Pro
16 results
Gemini 3 Pro
Undisclosed params · 13 results
Gemini-3.1-Pro
11 results
Gemma-3-27b
27B params · 11 results
Gemma 3 (27B, IT)
9 results
Gemma-2-27b-it
9 results
gemma-3-12b-it
9 results
§ 06 · Sources & freshness

Where these numbers come from.

arxiv
1
result
1 of 1 rows marked verified.