44 models tracked · known for gemma-3-1b-it, gemma-2-9b-it, gemma-2-2b-it
Claims drawn from cited facts, not live model generation.
Google DeepMind is an artificial intelligence research lab focused on cutting-edge breakthroughs. Its tracked blog publishes articles and stories about the lab's latest research and developments.
4 cited facts
Google DeepMind has shipped 44 models and has 100 recent blog announcements tracked in the graph. With no papers count offered, this mix of 44 models alongside 100 announcements signals an actively communicating lab focused on shipping product rather than a primarily research-driven footprint.
5 cited facts
401 benchmark scores across this lab’s models, tagged by evidence status.
| Model | Best score | Claim status |
|---|---|---|
| Gemini 4 Argon | 100 | unverified |
| Gemini 3.8 Flash | 98.89 | unverified |
| Gemini 3.1 Pro | 98 | independently reproduced |
| Gemini 3.7 Flash | 97.22 | independently reproduced |
| Gemini 2.5 Pro (May 2025) | 95.9 | independently reproduced |
| Gemini 2.5 Pro (Mar 2025) | 95.56 | independently reproduced |
| Gemini 3 Flash | 95.55 | independently reproduced |
| Gemini 3.5 Flash | 95.55 | independently reproduced |
| Gemini 3.6 Flash | 94.16 | independently reproduced |
| Gemini 2.5 Pro (Jun 2025) | 91.7 | independently reproduced |
| Gemini 3 Pro | 91.38 | independently reproduced |
| PaLM 2-L | 86.1 | self-reported |
| Gemini 1.5 Pro (May 2024) | 85.6 | independently reproduced |
| Gemma 2 9B | 84.9 | independently reproduced |
| Gemini 2.0 Pro | 83.46 | independently reproduced |
| Gemini 1.5 Pro (Sept 2024) | 82.53 | independently reproduced |
| Gemini 1.5 Flash (May 2024) | 82.4 | independently reproduced |
| Gemma 4 26B A4B | 82.2 | unverified |
| Gemini 2.0 Flash (Feb 2025) | 82.17 | independently reproduced |
| PaLM 2-M | 81.7 | self-reported |
| Gemini 3.1 Flash-Lite | 79.98 | independently reproduced |
| Gemma 3 27B | 79.9 | independently reproduced |
| Gemini 2.5 Flash (May 2025) | 77.8 | independently reproduced |
| Gemini 3.5 Flash-Lite | 77.78 | independently reproduced |
| Gemini 2.5 Flash (Apr 2025) | 76.5 | independently reproduced |
| Gemma 7B | 76.27 | self-reported |
| Gemini 1.5 Flash (Sep 2024) | 76 | independently reproduced |
| PaLM 2-S | 76 | self-reported |
| Gemini 2.0 Flash Thinking (Jan 2025) | 73.8 | independently reproduced |
| Gemma 4 31B IT | 73.31 | independently reproduced |
| Gemini 2.0 Flash (Dec 2024) | 72.93 | unverified |
| Gemma 2 27B | 67.6 | independently reproduced |
| Gemma 2B | 61.87 | unverified |
| Gemini 2.5 Flash (Jun 2025) | 60.88 | independently reproduced |
| Gemini 1.0 Pro | 60 | independently reproduced |
| Gemini 2.5 Flash-Lite (Jun 2025) | 47.2 | unverified |
| Gemini 2.5 Flash (Sep 2025) | 41.91 | unverified |
| Gemma 3 12B | 19.28 | unverified |
| Gemma 3 4B | 18.22 | unverified |
| gemma-3-1b-it | — | unverified |
| gemma-2-9b-it | — | unverified |
| gemma-2-2b-it | — | unverified |
| gemma-3-270m | — | unverified |
| gemma-3-270m-it | — | unverified |
Where a model from Google DeepMind holds the top score, with the benchmark’s integrity grade.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks, huggingface_models, lab_blog_rss