Up to date
Models › embeddinggemma-2 family
google Apache 2.0 out 41 days ago
744M parameters, text in, embeddings out.
Verified 2026-10-25 Rank #47 in Embeddings · score 89.9 method v1 Sources 3 Downloads 30d 206.9k · #248 of 619 tracked
Changed this week: 42 new quantized builds on the Hub (115 in all).2026-10-24 · 40 new quantized builds on the Hub (113 in all).2026-10-24
Picked by rule from 14 tracked pieces: the lab's own post, the most shared, a field report, the latest.
Google launches EmbeddingGemma 2, an open multimodal model expanding beyond text to unify code, images, video, and audio.
A GitHub project ports four jeffhub use cases to Google's EmbeddingGemma-2, replacing the original decider with cosine similarity scoring.
The 2 picks above, plus these 12. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.
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: 115 quantized builds.
Weights only, an estimate: leave room for the KV cache. Method
About 0.4 GB at INT4, 0.9 GB at FP8, 1.8 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
Third-party measurements only; author-reported numbers stay in Reference, marked as such.
67.53. Each tick is a model on this site with a measured MTEB row; author-reported numbers stay in Reference. Mean(task) x100 over 41 tasks. 41 tasks, split=test
Source · leaderboard · 2026-10-24
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2026-09-14), amber triangles are 8 dated articles (each links to it).
252 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Everything else, folded: the full tables, every source.
| Benchmark | Value | Setting / notes | Reported by | Date |
|---|---|---|---|---|
| measured MTEB · Embed text for search | 51.33 nDCG@10 x100 (mean of 10 tasks) | 10 tasks, split=test | leaderboard | 2026-10-24 |
| measured MTEB · Embed text for search | 67.53 mean(task) x100 over 41 tasks | 41 tasks, split=test | leaderboard | 2026-10-24 |
| measured MTEB Reranking · Rerank search results | 46.03 MAP x100 (mean of 2 tasks) | 2 tasks, split=test | leaderboard | 2026-10-24 |
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
Listed in Embeddings
| Pipeline tag | feature-extraction |
|---|---|
| Library | transformers |
| Architectures | EmbeddingGemma2Model |
| Model type | embedding_gemma2 |
| Params by dtype | BF16: 744,371,512 |
| Size band | S |
| MoE | no |
| License | Apache 2.0 (apache-2.0) |
| Hub snapshot | 2026-10-25 |
Weights-only memory for embeddinggemma-2: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 1.8 GB · Edge (<4 GB) |
|---|---|
| FP8 (1 byte/param) | 0.9 GB · Edge (<4 GB) |
| INT4 (0.5 bytes/param) | 0.4 GB · Edge (<4 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
quantized: 115finetune: 130adapter: 7
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| onnx-community/embeddinggemma-2-ONNX | quantized | — | — | — | apache-2.0 | onnx-community | 28 | 275.7k |
| unsloth/embeddinggemma-2-GGUF | quantized | GGUF | — | — | apache-2.0 | unsloth | 345 | 125k |
| litert-community/embeddinggemma-2-740m-litert-lm | finetune | — | — | — | apache-2.0 | litert-community | 36 | 130.7k |
| ggml-org/embeddinggemma-2-GGUF | quantized | GGUF | — | — | apache-2.0 | ggml-org | 54 | 90.8k |
| litert-community/embeddinggemma-2-text-270m-litert-lm | finetune | — | — | — | apache-2.0 | litert-community | 10 | 12.7k |
| litert-community/embeddinggemma-2-text-vision-440m-litert-lm | finetune | — | — | — | apache-2.0 | litert-community | 6 | 10.5k |
| unsloth/embeddinggemma-2 | finetune | — | 744M | S | apache-2.0 | unsloth | 12 | 5.9k |
| ngquocvinh/EmbeddingGemma-2-GGUF | quantized | GGUF | — | — | apache-2.0 | ngquocvinh | 1 | 6.4k |
| mlx-community/embeddinggemma-2-bf16 | finetune | -bf16 | 744M | S | apache-2.0 | mlx-community | 11 | 5.4k |
| abacusai/Smaug-AI-Image-Detector | finetune | — | 439M | S | apache-2.0 | abacusai | 6 | 5.2k |
| mlx-community/embeddinggemma-2-8bit | quantized | -8bit | 744M | S | apache-2.0 | mlx-community | 8 | 4.9k |
| mrm8488/embeddinggemma-2-ft-typed-decision | finetune | — | 272M | S | apache-2.0 | mrm8488 | 63 | 1.4k |
| qnguyen3/QUAN-Transcribe-1.0 | finetune | — | 102M | S | gemma | qnguyen3 | 64 | 544 |
| FluidInference/embeddinggemma-2-coreml | quantized | — | — | — | apache-2.0 | FluidInference | 27 | 1.6k |
| AtomicChat/embeddinggemma-2-GGUF | quantized | GGUF | — | — | apache-2.0 | AtomicChat | 2 | 2.7k |
| webmp3/Sakura-EmbeddingGemma-2-HighQuality-GGUF | quantized | GGUF | — | — | apache-2.0 | webmp3 | 1 | 1.9k |
| webmp3/Sakura-Gemma4-E2B-EmbeddingGemma2-DualMode-GGUF | quantized | GGUF | — | — | gemma | webmp3 | 1 | 1.9k |
| prithivMLmods/EmbeddingGemma-2-GGUF | quantized | GGUF | — | — | apache-2.0 | prithivMLmods | 1 | 1.3k |
| mlx-community/embeddinggemma-2-mxfp8 | quantized | — | 744M | S | apache-2.0 | mlx-community | 5 | 1.1k |
| BennyDaBall/EmbeddingGemma-2-NVFP4-GGUF | quantized | NVFP4 | — | — | apache-2.0 | BennyDaBall | 0 | 1.3k |
| mlx-community/embeddinggemma-2-4bit | quantized | -4bit | 744M | S | apache-2.0 | mlx-community | 3 | 951 |
| infosave/EmbeddingGemma-2-cmf | quantized | — | — | — | apache-2.0 | infosave | 1 | 790 |
| mlx-community/embeddinggemma-2-mxfp4 | quantized | mxfp4 | 744M | S | apache-2.0 | mlx-community | 3 | 666 |
| EverPaper/embeddinggemma-2-coreai | finetune | — | — | — | apache-2.0 | EverPaper | 16 | 0 |
| SargeDev/Solar-JEV-Gemma-embed-2-GGUF | quantized | GGUF | — | — | apache-2.0 | SargeDev | 0 | 787 |
| Fastiraz/embeddinggemma-2-GGUF | quantized | GGUF | — | — | apache-2.0 | Fastiraz | 0 | 697 |
| nativ-community/embeddinggemma-2-8bit | quantized | -8bit | 744M | S | apache-2.0 | nativ-community | 3 | 525 |
| webmp3/Sakura-EmbeddingGemma-2-AutoRound-GGUF | quantized | GGUF | — | — | gemma | webmp3 | 1 | 585 |
| webmp3/Sakura-Gemma4-E4B-EmbeddingGemma2-DualMode-GGUF | quantized | GGUF | — | — | gemma | webmp3 | 0 | 574 |
| PixelML/sea-asr-id | finetune | — | 112M | S | apache-2.0 | PixelML | 8 | 159 |
| mlx-community/embeddinggemma-2-nvfp4 | quantized | nvfp4 | 744M | S | apache-2.0 | mlx-community | 2 | 435 |
| mlx-community/embeddinggemma-2-5bit | quantized | — | 744M | S | apache-2.0 | mlx-community | 3 | 368 |
| yan2026/embeddinggemma-2-GGUF | quantized | GGUF | — | — | gemma | yan2026 | 0 | 494 |
| mlx-community/embeddinggemma-2-6bit | quantized | — | 744M | S | apache-2.0 | mlx-community | 3 | 337 |
| spectral-ai-lab/arbiter-272m | finetune | — | 272M | S | other | spectral-ai-lab | 5 | 180 |
| nativ-community/embeddinggemma-2-bf16 | finetune | -bf16 | 744M | S | apache-2.0 | nativ-community | 3 | 266 |
| pastel-labs/embeddinggemma-2-photo-anchor | finetune | — | — | — | cc-by-4.0 | pastel-labs | 8 | 0 |
| mlboydaisuke/EmbeddingGemma-2-Transcribe-LiteRT | finetune | — | — | — | apache-2.0 | mlboydaisuke | 6 | 95 |
| turintech/embeddinggemma-2-Artemis-GGUF | quantized | GGUF | — | — | apache-2.0 | turintech | 1 | 308 |
| HukukcuApp/tr-legal-embedding-gemma-2 | finetune | — | 271M | S | gemma | HukukcuApp | 4 | 138 |
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
Hub snapshot: 2026-10-25 · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-09 02:04 UTC · Last verified 2026-10-25
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.
