{ "data": { "similar": [ { "grade": "BB", "json": "https://www.anchorterminal.com/tools/qwen3-embedding-8b.json", "name": "Qwen3-Embedding-8B", "score": 74.7, "shared": [ "embed.text", "embed.multilingual", "embed.code", "model.open-weights" ], "slug": "qwen3-embedding-8b" }, { "grade": "BB", "json": "https://www.anchorterminal.com/tools/qwen3-embedding-4b.json", "name": "Qwen3-Embedding-4B", "score": 72.9, "shared": [ "embed.text", "embed.multilingual", "embed.code", "model.open-weights" ], "slug": "qwen3-embedding-4b" }, { "grade": "BB", "json": "https://www.anchorterminal.com/tools/gemini-embedding.json", "name": "Gemini Embedding", "score": 70.6, "shared": [ "embed.text", "embed.multimodal", "embed.code", "embed.multilingual" ], "slug": "gemini-embedding" }, { "grade": "B", "json": "https://www.anchorterminal.com/tools/qwen3-embedding-0-6b.json", "name": "Qwen3-Embedding-0.6B", "score": 63.5, "shared": [ "embed.text", "embed.multilingual", "embed.code", "model.open-weights" ], "slug": "qwen3-embedding-0-6b" }, { "grade": "B", "json": "https://www.anchorterminal.com/tools/qwen3-vl-embedding-8b.json", "name": "Qwen3-VL-Embedding-8B", "score": 62.1, "shared": [ "embed.text", "embed.multimodal", "embed.multilingual", "model.open-weights" ], "slug": "qwen3-vl-embedding-8b" }, { "grade": "C", "json": "https://www.anchorterminal.com/tools/jina-embeddings.json", "name": "Jina Embeddings and Reranker", "score": 61.2, "shared": [ "embed.text", "embed.multimodal", "embed.code", "embed.multilingual" ], "slug": "jina-embeddings" } ], "tool": { "slug": "embeddinggemma-2", "name": "EmbeddingGemma 2", "vendor": "Google DeepMind", "vendorUrl": "https://gemma.ai", "kind": "weights", "category": "embeddings", "summary": "EmbeddingGemma 2 is Google DeepMind's open-weight multimodal embedding model of 740M parameters, mapping text, code, images, video and audio into one 768-dimension vector space from up to 8,192 tokens, under Apache 2.0.", "url": "https://www.anchorterminal.com/tools/embeddinggemma-2", "markdownUrl": "https://www.anchorterminal.com/tools/embeddinggemma-2.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/embeddinggemma-2.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/embeddinggemma-2.json", "repo": "https://huggingface.co/google/embeddinggemma-2", "license": "Apache 2.0", "transports": [], "packages": [ { "registry": "huggingface", "name": "google/embeddinggemma-2" } ], "auth": "none", "authNotes": "Not gated on Hugging Face. The files download with no account and no terms to accept (gated false in the Hub's API, checked 2026-10-30).", "pricing": "free", "pricingNotes": "Free to download and use, commercial use included, under Apache 2.0, with the card adding that deployments must adhere to the Gemma Prohibited Use Policy. No inference provider serves it on Hugging Face. Running it takes little hardware, 1.53 GB of BF16 files, or about 0.5 GB in the 4-bit ONNX and LiteRT builds others publish (checked 2026-10-30).", "priceSummary": "Free", "where": "local", "x402": { "level": "no", "evidence": "Weights you download and run yourself, so there is nothing to pay per call and no payment protocol (checked 2026-10-30).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-10-30" }, "docsUrl": "https://ai.google.dev/gemma/docs/embeddinggemma", "llmsTxt": "https://gemma.ai/llms.txt", "capabilities": [ "embed.text", "embed.multimodal", "embed.multilingual", "embed.code", "model.open-weights" ], "tags": [ "open-weights", "model", "embeddings", "multimodal", "vision", "audio", "video", "multilingual", "matryoshka", "apache-2.0", "safetensors", "sentence-transformers", "self-hosted" ], "lastRelease": "2026-10-28", "graded": true, "anchor": { "graded": true, "score": 66.4, "grade": "B", "agentReady": false, "rank": 0, "ranked": false, "notRankedWhy": "An open-weight model, graded by the open-weight checklist and ranked among open-weight models, not against tools", "rankOf": 1239, "categoryRank": 7, "methodology": "0.7", "run": "2026-10-01", "scores": { "capability": 59, "interface": 70, "reach": 75, "risk": 60, "upkeep": 80 }, "pending": [ "performance", "tasks" ], "breakdown": [ { "key": "capability", "name": "Capability", "weight": 35, "effectiveWeight": 35, "score": 59, "points": 20.65, "reason": "From 9 benchmark results the publisher submitted (counted at half weight) in 2 areas, each scored as its share of the way from the benchmark's floor to the best result any model has on the same leaderboard. By area, Retrieval 63, Other MTEB task types 53. Publisher-submitted results count at half weight, and numbers that appear only on the publisher's card don't count.", "checked": "2026-10-30" }, { "key": "reach", "name": "Reach", "weight": 25, "effectiveWeight": 25, "score": 75, "points": 18.75, "reason": "Memory earns the full 30, since Google's BF16 files are 1.53 GB and 4-bit builds by others are about 0.5 GB. Transformers, vLLM, SGLang, llama.cpp and Ollama each run it in a released version (25). Builds score 0 because every quantised build sits in another organisation. The card applies the Gemma Prohibited Use Policy on top of Apache 2.0, so terms read as a custom licence with free commercial use (10). No gate (10). Hosted too scores 0, with no inference provider on Hugging Face.", "checked": "2026-10-30" }, { "key": "risk", "name": "Risk", "weight": 20, "effectiveWeight": 20, "score": 60, "points": 12, "reason": "File safety gets the full 25 for one safetensors file loaded by Transformers' own code. Provenance gets 15, Google's verified organisation and a checksum on every file. Safety evaluation is 0, the card has none. Agent guidance gets 10, since the card says there was no safety tuning and names the safeguards to add. Licence risk falls to 5, 5 off each for the use policy and Google's right to update it. The security programme gets 5 for Google's security.txt.", "checked": "2026-10-30" }, { "key": "interface", "name": "Agent interface", "weight": 10, "effectiveWeight": 10, "score": 70, "points": 7, "reason": "Read for an embedding model, the template, tool-calling, trained tool calling and structured-output lines stand for the prompt format and what it takes and returns. The template and named task prefixes ship in the repository (20), every modality's input and the output are documented (25), it was trained for task prefixes with one vector per input (10), and each vector size and how to get it is documented (15). Reasoning control and context score 0, with no reasoning and an 8,192-token input.", "checked": "2026-10-30" }, { "key": "upkeep", "name": "Documentation and upkeep", "weight": 10, "effectiveWeight": 10, "score": 80, "points": 8, "reason": "The card covers all seven items (20) and describes the training data by type with a cut-off (10). Google changed the repository two days ago (20). Google members answered 4 of 12 discussions (10). The earlier model keeps its own repository and gemma.ai dates both (10). Config, tokeniser, template and card agree, and no broken-file report waits (10).", "checked": "2026-10-30" } ], "assessment": { "date": "2026-10-30", "basis": "public evidence", "confidence": "medium", "notes": { "capability": "From 9 benchmark results the publisher submitted (counted at half weight) in 2 areas, each scored as its share of the way from the benchmark's floor to the best result any model has on the same leaderboard. By area, Retrieval 63, Other MTEB task types 53. Publisher-submitted results count at half weight, and numbers that appear only on the publisher's card don't count.", "interface": "Read for an embedding model, the template, tool-calling, trained tool calling and structured-output lines stand for the prompt format and what it takes and returns. The template and named task prefixes ship in the repository (20), every modality's input and the output are documented (25), it was trained for task prefixes with one vector per input (10), and each vector size and how to get it is documented (15). Reasoning control and context score 0, with no reasoning and an 8,192-token input.", "reach": "Memory earns the full 30, since Google's BF16 files are 1.53 GB and 4-bit builds by others are about 0.5 GB. Transformers, vLLM, SGLang, llama.cpp and Ollama each run it in a released version (25). Builds score 0 because every quantised build sits in another organisation. The card applies the Gemma Prohibited Use Policy on top of Apache 2.0, so terms read as a custom licence with free commercial use (10). No gate (10). Hosted too scores 0, with no inference provider on Hugging Face.", "risk": "File safety gets the full 25 for one safetensors file loaded by Transformers' own code. Provenance gets 15, Google's verified organisation and a checksum on every file. Safety evaluation is 0, the card has none. Agent guidance gets 10, since the card says there was no safety tuning and names the safeguards to add. Licence risk falls to 5, 5 off each for the use policy and Google's right to update it. The security programme gets 5 for Google's security.txt.", "upkeep": "The card covers all seven items (20) and describes the training data by type with a cut-off (10). Google changed the repository two days ago (20). Google members answered 4 of 12 discussions (10). The earlier model keeps its own repository and gemma.ai dates both (10). Config, tokeniser, template and card agree, and no broken-file report waits (10)." }, "sources": [ { "what": "model card, licence link, inference providers", "url": "https://huggingface.co/google/embeddinggemma-2", "seen": "2026-10-30" }, { "what": "not gated, created 14 September 2026, last modified 28 October 2026, 744,371,512 BF16 parameters, downloads", "url": "https://huggingface.co/api/models/google/embeddinggemma-2", "seen": "2026-10-30" }, { "what": "file list, sizes and checksums", "url": "https://huggingface.co/api/models/google/embeddinggemma-2/tree/main?recursive=true", "seen": "2026-10-30" }, { "what": "four commits, 6 to 28 October 2026", "url": "https://huggingface.co/api/models/google/embeddinggemma-2/commits/main", "seen": "2026-10-30" }, { "what": "21 discussions, 12 not pull requests, counted by author, member flag and date, text not read", "url": "https://huggingface.co/api/models/google/embeddinggemma-2/discussions", "seen": "2026-10-30" }, { "what": "architecture, sliding window, max positions", "url": "https://huggingface.co/google/embeddinggemma-2/raw/main/config.json", "seen": "2026-10-30" }, { "what": "task prefixes, Sentence Transformers 6.1.0 requirement", "url": "https://huggingface.co/google/embeddinggemma-2/raw/main/config_sentence_transformers.json", "seen": "2026-10-30" }, { "what": "Google's EmbeddingGemma repositories, no quantised EmbeddingGemma 2 build", "url": "https://huggingface.co/api/models?author=google\u0026search=embeddinggemma", "seen": "2026-10-30" }, { "what": "ONNX build for Transformers.js with WebGPU example", "url": "https://huggingface.co/onnx-community/embeddinggemma-2-ONNX", "seen": "2026-10-30" }, { "what": "LiteRT-LM int4 QAT builds", "url": "https://huggingface.co/litert-community/embeddinggemma-2-740m-litert-lm", "seen": "2026-10-30" }, { "what": "llama.cpp organisation's GGUF and mmproj", "url": "https://huggingface.co/ggml-org/embeddinggemma-2-GGUF", "seen": "2026-10-30" }, { "what": "Transformers release with embedding_gemma2, sliding window documented as a radius", "url": "https://raw.githubusercontent.com/huggingface/transformers/v5.19.0/src/transformers/models/embedding_gemma2/configuration_embedding_gemma2.py", "seen": "2026-10-30" }, { "what": "vLLM EmbeddingGemma2Model", "url": "https://raw.githubusercontent.com/vllm-project/vllm/v0.32.0/vllm/model_executor/models/registry.py", "seen": "2026-10-30" }, { "what": "SGLang EmbeddingGemma2Model", "url": "https://raw.githubusercontent.com/sgl-project/sglang/v0.5.22/python/sglang/srt/models/embedding_gemma2.py", "seen": "2026-10-30" }, { "what": "llama.cpp gemma-embedding2 architecture", "url": "https://raw.githubusercontent.com/ggml-org/llama.cpp/master/src/llama-arch.cpp", "seen": "2026-10-30" }, { "what": "Ollama EmbeddingGemma2Model in the MLX runner (partial clone)", "url": "https://github.com/ollama/ollama/tree/v0.40.13/mlxrunner/model/gemma4embedding", "seen": "2026-10-30" }, { "what": "Transformers.js embedding_gemma2 (partial clone)", "url": "https://github.com/huggingface/transformers.js/tree/4.3.2/packages/transformers/src/models/embedding_gemma2", "seen": "2026-10-30" }, { "what": "MTEB rows for EmbeddingGemma 2 and 300M", "url": "https://mteb-leaderboard-backend.hf.space/v1/benchmarks/MTEB%28Multilingual%2C%20v2%29/scores", "seen": "2026-10-30" }, { "what": "dated releases, per the 21 October 2026 check", "url": "https://gemma.ai/models/releases.md", "seen": "2026-10-30" }, { "what": "Google security.txt, per the 21 October 2026 check", "url": "https://www.google.com/.well-known/security.txt", "seen": "2026-10-30" }, { "what": "Independent benchmark result used for Capability", "url": "https://huggingface.co/spaces/mteb/leaderboard", "seen": "2026-10-30" } ], "openQuestions": [ "Terms and licence risk read the card's line that deployments must adhere to the Gemma Prohibited Use Policy as a term added to Apache 2.0. Read as a request, terms is 15 and licence risk 15, Reach 80 and Risk 70", "Unchecked in this session, ai.google.dev (documentation, licence and policy pages), blog.google, gemma.ai and arXiv, since the fetcher refused them; their facts come from the 21 October 2026 checks of EmbeddingGemma 300M and Gemma 4", "If the card's Prohibited Use Policy link still shows only the moved-documentation note it showed on 21 October, a strict reading takes licence risk to 0", "Builds would be 10 if a Google page links litert-community's int4 builds as Google's own; none read here does", "Unchecked, LM Studio's catalogue and Ollama's library for an EmbeddingGemma 2 entry; neither changes the runtimes score", "Unchecked today, whether the Gemini API or Vertex AI serves EmbeddingGemma 2", "Agent guidance takes the 10 band on the card's statement that there was no safety tuning plus named safeguards; EmbeddingGemma 300M's card, with general cautions only, got 5" ] }, "negative": 0, "verdict": "A small multimodal embedding cousin that puts text, code, images, video and audio in one 768-dimension space, with task prefixes and vectors that truncate to 128. Five common runtimes run it and the BF16 files are 1.53 GB, with a browser build for Transformers.js. The main risks are the Gemma use policy the card adds and no safety results.", "bestFor": "Agents that need one small local embedding model for text, code, images, video and audio search", "strengths": [ "Apache 2.0, not gated, with safetensors only in Google's verified organisation and a checksum on every file", "Text, images, video and audio in one 8,192-token input and one 768-dimension vector, truncatable to 512, 256 or 128", "Runs in Transformers 5.19.0 with Sentence Transformers 6.1.0, vLLM 0.32.0, SGLang 0.5.22, llama.cpp and Ollama 0.40.13", "Vision and audio encoders load only when needed, so text alone is 270M parameters", "A card that documents every modality's token cost, the task prefixes and a January 2025 cut-off" ], "weaknesses": [ "The card requires deployments to follow the Gemma Prohibited Use Policy, which Google can update", "Google publishes no quantised build of its own; the 4-bit and GGUF builds are other organisations'", "An 8,192-token input shared by every modality, about 29 images or 327 seconds of audio", "No safety results, and no inference provider hosts it", "MTEB's rows for it sit below the card's figures on every tab read, and below EmbeddingGemma 300M on the two text tabs" ], "agentNotes": [ "Install sentence-transformers 6.1.0 or later and transformers 5.19.0 or later; older Sentence Transformers ignore the order of multimodal dict inputs", "Use prompt_name SearchQuery for queries and Document for untitled documents; write `title: {title} | text: {content}` yourself for titled ones, and pass images, video and audio with no prefix", "For text only, load with config_kwargs {\"vision_config\": None, \"audio_config\": None} to skip 470M parameters of encoders", "Run in bfloat16 or float32, never float16, which returns NaN or degraded vectors without an error", "Pass truncate_dim and normalize_embeddings=True for smaller vectors, and keep queries and documents at the same size", "Supply audio as mono 16 kHz; video is sampled at 1 frame a second by default" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 0, "avgRating": 0, "history": [ { "basis": "public evidence", "confidence": "medium", "grade": "B", "methodology": "0.7", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 66.4 } ], "editorialScores": { "capability": 59, "interface": 70, "reach": 75, "risk": 60, "upkeep": 80 }, "provenanceScore": 70, "checked": { "categories": { "capability": "2026-10-30", "interface": "2026-10-30", "reach": "2026-10-30", "risk": "2026-10-30", "upkeep": "2026-10-30" }, "last": "2026-10-30", "oldest": "2026-10-30", "stale": false, "staleAfterDays": 90 }, "scoreHistory": [ { "date": "2026-10-30", "grade": "B", "kind": "listed", "method": "0.7", "score": 66.4 } ] }, "connect": { "install": "pip install -U \"sentence-transformers\u003e=6.1.0\" \"transformers\u003e=5.19.0\"\npython -c \"from sentence_transformers import SentenceTransformer; m=SentenceTransformer('google/embeddinggemma-2', config_kwargs={'vision_config': None, 'audio_config': None}); print(m.encode('Which planet is red?', prompt_name='SearchQuery').shape)\"" }, "letme": { "capability": "https://letme.dev/embed.text", "tool": "https://letme.dev/embeddinggemma-2" }, "sameCompany": [ "embeddinggemma-300m", "diffusiongemma-26b-a4b-it", "gemma-4-12b-it", "gemma-4-26b-a4b-it", "gemma-4-31b-it", "gemma-4-e2b-it", "gemma-4-e4b-it" ], "weights": { "trainedBy": "Google DeepMind", "builtOn": "Gemma 4 architecture (the card says it builds on Gemma 4's advancements; no base checkpoint named)", "params": "740M", "context": 8192, "dimensions": 768, "goodFor": "Small multimodal embeddings for search, retrieval, classification and clustering across text, code, images, video and audio in one vector space, on laptops, phones or in a browser.", "runtimes": [ "Transformers", "vLLM", "SGLang", "llama.cpp", "Ollama" ], "builds": [ { "name": "embeddinggemma-2", "url": "https://huggingface.co/google/embeddinggemma-2", "who": "publisher", "precision": "bf16", "sizeGB": 1.53, "memoryGB": 8, "format": "safetensors", "runsOn": [ "Transformers", "vLLM", "SGLang" ], "note": "All three encoders. Text only loads 270M of the 740M parameters" }, { "name": "ggml-org embeddinggemma-2-Q8_0", "url": "https://huggingface.co/ggml-org/embeddinggemma-2-GGUF", "who": "llama.cpp", "precision": "q8_0", "sizeGB": 0.86, "memoryGB": 8, "format": "gguf", "runsOn": [ "llama.cpp" ], "note": "llama.cpp's organisation converted it automatically from Google's 6 October commit. The text model is 0.31 GB and the multimodal projector 0.55 GB; BF16 files are there too" }, { "name": "onnx-community embeddinggemma-2-ONNX q4", "url": "https://huggingface.co/onnx-community/embeddinggemma-2-ONNX", "who": "onnx-community", "precision": "q4", "sizeGB": 0.5, "memoryGB": 8, "format": "onnx", "runsOn": [ "Transformers.js" ], "note": "Hugging Face's ONNX community build for Transformers.js, WebGPU, WASM or Node. q4 text model 0.17 GB, vision encoder 0.11 GB, audio encoder 0.19 GB, tokeniser 0.03 GB; fp32, fp16, q4f16 and 8-bit files too" }, { "name": "litert-community embeddinggemma-2-740m-litert-lm", "url": "https://huggingface.co/litert-community/embeddinggemma-2-740m-litert-lm", "who": "litert-community", "precision": "int4 (QAT, int8 vision, mixed audio)", "sizeGB": 0.48, "memoryGB": 8, "format": "litertlm", "runsOn": [ "LiteRT-LM" ], "note": "The LiteRT Community organisation's on-device build, linked to Google AI Edge's apps and demos; text-only (165 MB) and text and vision (388 MB) builds sit in their own repositories. Not counted as Google's own, since no first-hand Google page read here links it" } ], "run": "python -c \"from sentence_transformers import SentenceTransformer; SentenceTransformer('google/embeddinggemma-2')\"", "runBuild": "embeddinggemma-2", "checked": "2026-10-30", "minMemoryGB": 8 }, "family": { "lab": "Google (Gemma)", "family": "Gemma 4", "url": "https://www.anchorterminal.com/open-weights/#google", "others": [ { "name": "gemma-4-26B-A4B", "url": "https://huggingface.co/google/gemma-4-26B-A4B", "params": "25.2B total, 3.8B active" }, { "name": "gemma-4-26B-A4B-it", "url": "https://huggingface.co/google/gemma-4-26B-A4B-it", "params": "25.2B total, 3.8B active", "listing": "https://www.anchorterminal.com/tools/gemma-4-26b-a4b-it" }, { "name": "gemma-4-31B", "url": "https://huggingface.co/google/gemma-4-31B", "params": "30.7B" }, { "name": "gemma-4-31B-it", "url": "https://huggingface.co/google/gemma-4-31B-it", "params": "30.7B", "listing": "https://www.anchorterminal.com/tools/gemma-4-31b-it" }, { "name": "gemma-4-E2B", "url": "https://huggingface.co/google/gemma-4-E2B", "params": "2.3B effective (5.1B with embeddings)" }, { "name": "gemma-4-E2B-it", "url": "https://huggingface.co/google/gemma-4-E2B-it", "params": "2.3B effective (5.1B with embeddings)", "listing": "https://www.anchorterminal.com/tools/gemma-4-e2b-it" }, { "name": "gemma-4-E4B", "url": "https://huggingface.co/google/gemma-4-E4B", "params": "4.5B effective (8B with embeddings)" }, { "name": "gemma-4-E4B-it", "url": "https://huggingface.co/google/gemma-4-E4B-it", "params": "4.5B effective (8B with embeddings)", "listing": "https://www.anchorterminal.com/tools/gemma-4-e4b-it" }, { "name": "gemma-4-26B-A4B-it-assistant", "url": "https://huggingface.co/google/gemma-4-26B-A4B-it-assistant", "params": "420M (file metadata on Hugging Face; drafter)" }, { "name": "gemma-4-31B-it-assistant", "url": "https://huggingface.co/google/gemma-4-31B-it-assistant", "params": "470M (file metadata on Hugging Face; drafter)" }, { "name": "gemma-4-E2B-it-assistant", "url": "https://huggingface.co/google/gemma-4-E2B-it-assistant", "params": "78M (file metadata on Hugging Face; drafter)" }, { "name": "gemma-4-E4B-it-assistant", "url": "https://huggingface.co/google/gemma-4-E4B-it-assistant", "params": "79M (file metadata on Hugging Face; drafter)" }, { "name": "gemma-4-12B", "url": "https://huggingface.co/google/gemma-4-12B", "params": "11.95B" }, { "name": "gemma-4-12B-it", "url": "https://huggingface.co/google/gemma-4-12B-it", "params": "11.95B", "listing": "https://www.anchorterminal.com/tools/gemma-4-12b-it" }, { "name": "gemma-4-12B-it-assistant", "url": "https://huggingface.co/google/gemma-4-12B-it-assistant", "params": "423M (file metadata on Hugging Face; drafter)" }, { "name": "diffusiongemma-26B-A4B-it", "url": "https://huggingface.co/google/diffusiongemma-26B-A4B-it", "params": "25.2B total, 3.8B active", "listing": "https://www.anchorterminal.com/tools/diffusiongemma-26b-a4b-it" } ], "olderCount": 171 }, "notable": [ "338,269 downloads in the last 30 days and 3,026 likes on Hugging Face on 30 October 2026 (https://huggingface.co/api/models/google/embeddinggemma-2)", "The MTEB leaderboard lists it with every task scored on MTEB(Multilingual, v2) at 60.52 mean task score, MTEB(eng, v2) at 67.53, MTEB(Code, v1) at 77.88, MIEB(lite) at 63.52 and MAEB(v1) at 49.31. Each is below the figure on Google's card (61.36, 68.46, 78.68, 65.67 and 49.87), and on the two text tabs below EmbeddingGemma 300M's own rows (61.14 and 69.67). MTEB doesn't say who ran them (https://mteb-leaderboard-backend.hf.space/v1/benchmarks/MTEB%28Multilingual%2C%20v2%29/scores)", "Google's card claims about a 14 per cent improvement on code tasks over EmbeddingGemma 300M. The MTEB(Code, v1) rows give 77.88 against 68.76, 13 per cent (https://huggingface.co/google/embeddinggemma-2)", "On 27 October 2026 Google raised the card's MIEB(lite) figure from 64.64 to 65.67 and its MAEB figure from 49.39 to 49.87, with no note on the card (https://huggingface.co/api/models/google/embeddinggemma-2/commits/main)", "The browser claim from the launch has a published build. onnx-community/embeddinggemma-2-ONNX holds fp32, fp16, 8-bit, q4 and q4f16 ONNX files for Transformers.js with a WebGPU example, and Transformers.js 4.3.2 ships an embedding_gemma2 model. Hugging Face's ONNX community published it, not Google, and we haven't run it (https://huggingface.co/onnx-community/embeddinggemma-2-ONNX)", "EmbeddingGemma 300M, the earlier text-only model under the Gemma Terms of Use, stays published in its own repository and is listed separately" ], "area": "models", "details": [ { "label": "Surface graded", "value": "The open weights at huggingface.co/google/embeddinggemma-2, read from the card, config files, Sentence Transformers configs, processor config, template, the Hub's metadata, file list, commits and discussion list on 30 October 2026, with the runtimes' own code and the MTEB leaderboard's data. Google's documentation, launch blog, technical report, gemma.ai and the licence pages couldn't be fetched in this session; what this listing says about them comes from the EmbeddingGemma 300M and Gemma 4 checks of 21 October 2026. Nothing was downloaded or run" }, { "label": "Model", "value": "A bidirectional encoder built on the Gemma 4 architecture (EmbeddingGemma2Model), 24 layers with five sliding-window layers to each global one, mean pooling and a 512 to 768 projection. 740M parameters per Google, 744,371,512 BF16 in the safetensors file: a 270M text model (130M transformer and 140M embeddings) with a 170M vision encoder and a 300M audio encoder that can be left out at load time" }, { "label": "Input and output", "value": "Text, images, video and audio, alone or interleaved with \u003c|image|\u003e, \u003c|video|\u003e and \u003c|audio|\u003e placeholders, share one 8,192-token budget. Text costs a token per subword; an image 280 tokens by default (a budget of 70 to 1,120), about 29 images; video 140 tokens a frame at 1 frame per second, about 58 frames; audio, mono at 16 kHz, 25 tokens a second, about 327 seconds. Out comes one mean-pooled, L2-normalised 768-dimension vector, truncatable to 512, 256 or 128 and re-normalised. Text takes a task prefix, eight kinds named in config_sentence_transformers.json (SearchQuery, Document, QuestionAnswering, FactChecking, CodeRetrieval, Classification, Clustering, SentenceSimilarity, with aliases); media take none. bfloat16 or float32, never float16" }, { "label": "Files", "value": "model.safetensors (1.49 GB, BF16), the tokeniser (32 MB tokenizer.json and a SentencePiece model), config, processor, template and Sentence Transformers configs, 1.53 GB in all. No pickle files and no Python. Every file carries a checksum in the Hub's file list" }, { "label": "Licence", "value": "Apache 2.0 in the card's metadata, with the card's licence link pointing to Google's Gemma 4 licence page, which on 21 October 2026 redirected to Google's plain Apache 2.0 page. No LICENSE file in the repository. The card's Ethics and Safety section adds that deployments must adhere to the Gemma Prohibited Use Policy, which Google says it may update; Gemma 4's cards carry no such line" }, { "label": "Runtimes", "value": "Transformers 5.19.0 and later (embedding_gemma2) with Sentence Transformers 6.1.0 or later, vLLM 0.32.0 (EmbeddingGemma2Model in its registry), SGLang 0.5.22 (embedding_gemma2.py), llama.cpp (gemma-embedding2 architecture, with ggml-org's GGUF and mmproj files), Ollama 0.40.13 (its MLX runner's gemma4embedding model and import code). Transformers.js 4.3.2 runs the ONNX build in the browser. LM Studio's catalogue and Ollama's library weren't checked" }, { "label": "Builds", "value": "Google publishes BF16 only in its own organisation. Others publish quantised builds: onnx-community's ONNX (q4 about 0.5 GB with all three encoders), litert-community's LiteRT-LM int4 QAT builds (165 MB text, 388 MB text and vision, 485 MB full, plus NPU builds), ggml-org's BF16 and Q8_0 GGUF with a separate multimodal projector, unsloth's GGUF and mlx-community's MLX" }, { "label": "Training data", "value": "Web documents in over 140 languages, code, images, video, audio and paired cross-modal samples, with a January 2025 cut-off, filtered for CSAM, personal and sensitive data, and quality. No sources, mix or sizes" }, { "label": "Hardware", "value": "The card names mobile devices and laptops as targets and gives text-only, text and image, text and audio and full loading sizes (270M, 440M, 570M, 740M parameters) but no memory figure. The BF16 files are 1.53 GB" } ], "provenance": { "legalEntity": "Google LLC (RDAP registrant of gemma.ai; the card names Google DeepMind as author)", "domain": "gemma.ai", "domainRegistered": "2017-12-16", "endpointOnVendorDomain": null, "terms": "https://ai.google.dev/gemma/docs/gemma_4_license", "privacy": "", "statusPage": "", "changelog": "https://gemma.ai/models/releases.md", "securityTxt": "none", "checked": "2026-10-30", "notes": [ "The registrant, registration date, releases page and security.txt findings are carried from the embeddinggemma-300m and gemma-4 facts checked on 21 October 2026; gemma.ai and ai.google.dev couldn't be fetched in this session.", "Nothing is hosted, so the licence stands in for terms of service and no privacy policy is scored.", "The repository was created privately on 14 September 2026 and published on 6 October 2026 (the super-squash commit, MTEB's release date and Transformers' contribution note agree). The card last changed on 28 October 2026.", "gemma.ai/.well-known/security.txt answered 404 on 21 October 2026. Google's company-wide security.txt points to its vulnerability reward programme." ], "score": 70, "checks": [ { "check": "Legal entity named", "value": "Google LLC (RDAP registrant of gemma.ai; the card names Google DeepMind as author)", "points": 20, "max": 20, "state": "ok" }, { "check": "Domain age", "value": "gemma.ai, registered 2017-12-16 (8 years)", "points": 11, "max": 15, "state": "part" }, { "check": "Endpoint on the vendor's domain", "value": "no hosted endpoint", "points": 0, "max": 0, "state": "na" }, { "check": "Terms of service", "value": "read, states 3 of the 7 things a reader expects", "points": 6.6, "max": 10, "state": "part" }, { "check": "Privacy policy", "value": "nothing hosted, not scored", "points": 0, "max": 0, "state": "na" }, { "check": "Status page", "value": "not found", "points": 0, "max": 10, "state": "no" }, { "check": "Changelog", "value": "published", "points": 10, "max": 10, "state": "ok" }, { "check": "security.txt", "value": "not found", "points": 0, "max": 10, "state": "no" }, { "check": "People named", "value": "6 named with their role, 2 with a public profile", "points": 10, "max": 10, "state": "ok" }, { "check": "Support route", "value": "help pages, a community", "points": 5, "max": 5, "state": "ok" } ], "policies": [ { "kind": "terms", "url": "https://ai.google.dev/gemma/docs/gemma_4_license", "state": "read", "readAt": "2026-10-30", "statedDate": "2026-04-01", "words": 1998, "points": 6.6, "max": 10, "expected": [ { "key": "terms.date", "label": "Gives the date it was last updated", "found": true, "quote": "Last updated 2026-04-01 UTC.", "says": "Last updated 2026-04-01" }, { "key": "terms.law", "label": "Names the governing law or courts", "found": false }, { "key": "terms.liability", "label": "States a limit on its liability", "found": true, "quote": "Work (including but not limited to damages for loss of goodwill," }, { "key": "terms.termination", "label": "Says how the agreement or account can be ended", "found": false }, { "key": "terms.changes", "label": "Says how changes to the terms are announced", "found": false }, { "key": "terms.use", "label": "Lists what users may not do", "found": true, "quote": "Prohibited use" }, { "key": "terms.sla", "label": "Refers to a service level or uptime commitment", "found": false } ] }, { "kind": "privacy", "url": "", "state": "none-found", "points": 0, "max": 10 } ] }, "humans": { "slug": "embeddinggemma-2", "checked": "2026-10-30", "company": "Google DeepMind", "people": [ { "name": "Demis Hassabis", "role": "Co-founder and CEO, Google DeepMind", "founder": true, "bio": "Demis Hassabis co-founded DeepMind in 2010 with Shane Legg and Mustafa Suleyman and leads Google DeepMind. He is also chief scientist of Alphabet and co-founder and CEO of Isomorphic Labs. In 2024 he shared the Nobel Prize in Chemistry with John Jumper for AI work on protein structure prediction.", "bioSources": [ "https://deepmind.google/about/", "https://en.wikipedia.org/wiki/Demis_Hassabis", "https://en.wikipedia.org/wiki/Google_DeepMind" ], "page": "/people/demis-hassabis/" }, { "name": "Shane Legg", "role": "Co-founder and Chief AGI Scientist, Google DeepMind", "founder": true, "bio": "Shane Legg co-founded DeepMind with Demis Hassabis and Mustafa Suleyman and works at Google DeepMind as chief AGI scientist. In the early 2000s he helped bring the term artificial general intelligence into use with Ben Goertzel, and his doctoral thesis was supervised by Marcus Hutter.", "bioSources": [ "https://en.wikipedia.org/wiki/Shane_Legg", "https://en.wikipedia.org/wiki/Google_DeepMind" ], "page": "/people/shane-legg/" }, { "name": "Mustafa Suleyman", "role": "Co-founder and former head of applied AI, DeepMind", "founder": true, "bio": "Mustafa Suleyman co-founded DeepMind and was its head of applied AI. After leaving DeepMind he co-founded Inflection AI in 2022, and he is now CEO of Microsoft AI.", "bioSources": [ "https://en.wikipedia.org/wiki/Mustafa_Suleyman", "https://en.wikipedia.org/wiki/Google_DeepMind" ], "page": "/people/mustafa-suleyman/" }, { "name": "Gus Martins", "role": "Gemma Product Manager", "linkedin": "https://uk.linkedin.com/in/lgusmartins", "x": "https://x.com/gusthema", "page": "/people/gus-martins/" }, { "name": "Henrique Schechter Vera", "role": "Lead author, EmbeddingGemma 2 technical report", "page": "/people/henrique-schechter-vera/" }, { "name": "Sahil Dua", "role": "Author, EmbeddingGemma 2 technical report", "x": "https://x.com/sahildua2305", "page": "/people/sahil-dua/" } ], "peopleNote": "The founders, Gus Martins and Sahil Dua's X account are carried from data/humans/embeddinggemma-300m.json (checked 28 October 2026), the same company. EmbeddingGemma 2's model card, read on 30 October 2026, cites its technical report (arXiv 2610.37364) as by Henrique Schechter Vera, Sahil Dua and the EmbeddingGemma team; both also appear among the Hugging Face accounts in the repository's first commit. The launch blog, the report itself and Google's documentation couldn't be fetched in this session, so no other role is listed and no account is given for Henrique Schechter Vera. ownership is carried from the 300M file (subsidiary of Alphabet Inc., checked 28 October 2026); a subsidiary gets no investors.", "companySocial": { "linkedin": "https://www.linkedin.com/company/googledeepmind/", "x": "https://x.com/googlegemma", "github": "https://github.com/google-gemma" }, "support": { "urls": [ "https://huggingface.co/google/embeddinggemma-2/discussions", "https://ai.google.dev/gemma/docs/embeddinggemma" ], "community": [ { "label": "Model discussions", "url": "https://huggingface.co/google/embeddinggemma-2/discussions", "kind": "link" }, { "label": "Google developer Discord #gemma", "url": "https://discord.gg/google-dev-community", "kind": "discord" } ] }, "sources": [ "https://huggingface.co/google/embeddinggemma-2", "https://huggingface.co/api/models/google/embeddinggemma-2/commits/main", "https://arxiv.org/abs/2610.37364", "https://deepmind.google/about/", "https://en.wikipedia.org/wiki/Google_DeepMind", "https://en.wikipedia.org/wiki/Demis_Hassabis", "https://en.wikipedia.org/wiki/Shane_Legg", "https://en.wikipedia.org/wiki/Mustafa_Suleyman", "https://uk.linkedin.com/in/lgusmartins", "https://x.com/gusthema", "https://x.com/sahildua2305", "https://ai.google.dev/gemma/docs/discord" ], "ownership": { "status": "subsidiary", "parent": "Alphabet Inc.", "checked": "2026-10-28", "sources": [ "https://deepmind.google/about/", "https://en.wikipedia.org/wiki/Google_DeepMind" ] } }, "pageJsonUrl": "https://www.anchorterminal.com/tools/embeddinggemma-2.json" }, "verify": { "accepts": "a page on gemma.ai or one of its subdomains", "badgeUrl": "https://www.anchorterminal.com/badges/embeddinggemma-2.svg", "body": { "slug": "embeddinggemma-2", "url": "the page with the badge or the link" }, "docs": "https://www.anchorterminal.com/builders/#verify", "effect": "none, it never changes a grade, rank or review", "endpoint": "https://www.anchorterminal.com/api/v1/verify", "listingUrl": "https://www.anchorterminal.com/tools/embeddinggemma-2", "mcpTool": "verify_listing", "recheck": "weekly; two failed checks in a row and it lapses, a later pass restores it", "snippets": { "html": "\u003ca href=\"https://www.anchorterminal.com/tools/embeddinggemma-2\"\u003e\u003cimg src=\"https://www.anchorterminal.com/badges/embeddinggemma-2.svg\" alt=\"EmbeddingGemma 2 on Anchor Terminal\" height=\"20\"\u003e\u003c/a\u003e", "markdown": "[![EmbeddingGemma 2 on Anchor Terminal](https://www.anchorterminal.com/badges/embeddinggemma-2.svg)](https://www.anchorterminal.com/tools/embeddinggemma-2)", "link": "\u003ca href=\"https://www.anchorterminal.com/tools/embeddinggemma-2\"\u003eEmbeddingGemma 2 on Anchor Terminal\u003c/a\u003e" } } }, "kind": "anchor.page", "links": { "api": "https://www.anchorterminal.com/api/v1/index.json", "html": "https://www.anchorterminal.com/tools/embeddinggemma-2", "json": "https://www.anchorterminal.com/tools/embeddinggemma-2.json", "llms": "https://www.anchorterminal.com/llms.txt", "markdown": "https://www.anchorterminal.com/tools/embeddinggemma-2.md", "mcp": "https://www.anchorterminal.com/mcp", "pricing": "https://www.anchorterminal.com/tools/embeddinggemma-2#pricing", "reviews": "https://www.anchorterminal.com/tools/embeddinggemma-2#reviews", "slim": "https://www.anchorterminal.com/tools/embeddinggemma-2.min.md" }, "markdown": "## Overview\n\n**Grade B · 66.4/100 · rank #16 of 117 open-weight models · #7 in Embeddings \u0026 rerankers · not agent-ready · confidence medium**\n\n\nMore from Google DeepMind, listed separately because each is its own product: [EmbeddingGemma 300M](https://www.anchorterminal.com/tools/embeddinggemma-300m.md) (Embeddings \u0026 rerankers), [DiffusionGemma 26B A4B IT](https://www.anchorterminal.com/tools/diffusiongemma-26b-a4b-it.md) (Open-weight models), [Gemma 4 12B IT](https://www.anchorterminal.com/tools/gemma-4-12b-it.md) (Open-weight models), [Gemma 4 26B A4B IT](https://www.anchorterminal.com/tools/gemma-4-26b-a4b-it.md) (Open-weight models), [Gemma 4 31B IT](https://www.anchorterminal.com/tools/gemma-4-31b-it.md) (Open-weight models), [Gemma 4 E2B IT](https://www.anchorterminal.com/tools/gemma-4-e2b-it.md) (Open-weight models), [Gemma 4 E4B IT](https://www.anchorterminal.com/tools/gemma-4-e4b-it.md) (Open-weight models).\n\nTrained by Google (Gemma), in the Gemma 4 family, with [gemma-4-26B-A4B](https://huggingface.co/google/gemma-4-26B-A4B), [gemma-4-26B-A4B-it](https://www.anchorterminal.com/tools/gemma-4-26b-a4b-it.md), [gemma-4-31B](https://huggingface.co/google/gemma-4-31B), [gemma-4-31B-it](https://www.anchorterminal.com/tools/gemma-4-31b-it.md), [gemma-4-E2B](https://huggingface.co/google/gemma-4-E2B), [gemma-4-E2B-it](https://www.anchorterminal.com/tools/gemma-4-e2b-it.md), [gemma-4-E4B](https://huggingface.co/google/gemma-4-E4B), [gemma-4-E4B-it](https://www.anchorterminal.com/tools/gemma-4-e4b-it.md), [gemma-4-26B-A4B-it-assistant](https://huggingface.co/google/gemma-4-26B-A4B-it-assistant), [gemma-4-31B-it-assistant](https://huggingface.co/google/gemma-4-31B-it-assistant), [gemma-4-E2B-it-assistant](https://huggingface.co/google/gemma-4-E2B-it-assistant), [gemma-4-E4B-it-assistant](https://huggingface.co/google/gemma-4-E4B-it-assistant), [gemma-4-12B](https://huggingface.co/google/gemma-4-12B), [gemma-4-12B-it](https://www.anchorterminal.com/tools/gemma-4-12b-it.md), [gemma-4-12B-it-assistant](https://huggingface.co/google/gemma-4-12B-it-assistant), [diffusiongemma-26B-A4B-it](https://www.anchorterminal.com/tools/diffusiongemma-26b-a4b-it.md). Every model from this lab and its older versions: https://www.anchorterminal.com/open-weights/#google.\n\n## Assessment\n\nA small multimodal embedding cousin that puts text, code, images, video and audio in one 768-dimension space, with task prefixes and vectors that truncate to 128. Five common runtimes run it and the BF16 files are 1.53 GB, with a browser build for Transformers.js. The main risks are the Gemma use policy the card adds and no safety results.\n\n## Facts\n\n| Field | Value |\n| --- | --- |\n| Vendor | Google DeepMind (https://gemma.ai) |\n| Kind | Open-weight model |\n| Category | Embeddings \u0026 rerankers (https://www.anchorterminal.com/categories/embeddings) |\n| Auth | None · Not gated on Hugging Face. The files download with no account and no terms to accept (gated false in the Hub's API, checked 2026-10-30). |\n| Pricing | Free (Free) · Free to download and use, commercial use included, under Apache 2.0, with the card adding that deployments must adhere to the Gemma Prohibited Use Policy. No inference provider serves it on Hugging Face. Running it takes little hardware, 1.53 GB of BF16 files, or about 0.5 GB in the 4-bit ONNX and LiteRT builds others publish (checked 2026-10-30). |\n| x402 | No · Weights you download and run yourself, so there is nothing to pay per call and no payment protocol (checked 2026-10-30). |\n| Licence | Apache 2.0 |\n| Packages | huggingface: `google/embeddinggemma-2` |\n| Source | https://huggingface.co/google/embeddinggemma-2 |\n| Docs | https://ai.google.dev/gemma/docs/embeddinggemma |\n| llms.txt | https://gemma.ai/llms.txt |\n| Last release | 2026-10-28 |\n| Surface graded | The open weights at huggingface.co/google/embeddinggemma-2, read from the card, config files, Sentence Transformers configs, processor config, template, the Hub's metadata, file list, commits and discussion list on 30 October 2026, with the runtimes' own code and the MTEB leaderboard's data. Google's documentation, launch blog, technical report, gemma.ai and the licence pages couldn't be fetched in this session; what this listing says about them comes from the EmbeddingGemma 300M and Gemma 4 checks of 21 October 2026. Nothing was downloaded or run |\n| Model | A bidirectional encoder built on the Gemma 4 architecture (EmbeddingGemma2Model), 24 layers with five sliding-window layers to each global one, mean pooling and a 512 to 768 projection. 740M parameters per Google, 744,371,512 BF16 in the safetensors file: a 270M text model (130M transformer and 140M embeddings) with a 170M vision encoder and a 300M audio encoder that can be left out at load time |\n| Input and output | Text, images, video and audio, alone or interleaved with \u003c\\|image\\|\u003e, \u003c\\|video\\|\u003e and \u003c\\|audio\\|\u003e placeholders, share one 8,192-token budget. Text costs a token per subword; an image 280 tokens by default (a budget of 70 to 1,120), about 29 images; video 140 tokens a frame at 1 frame per second, about 58 frames; audio, mono at 16 kHz, 25 tokens a second, about 327 seconds. Out comes one mean-pooled, L2-normalised 768-dimension vector, truncatable to 512, 256 or 128 and re-normalised. Text takes a task prefix, eight kinds named in config_sentence_transformers.json (SearchQuery, Document, QuestionAnswering, FactChecking, CodeRetrieval, Classification, Clustering, SentenceSimilarity, with aliases); media take none. bfloat16 or float32, never float16 |\n| Files | model.safetensors (1.49 GB, BF16), the tokeniser (32 MB tokenizer.json and a SentencePiece model), config, processor, template and Sentence Transformers configs, 1.53 GB in all. No pickle files and no Python. Every file carries a checksum in the Hub's file list |\n| Licence | Apache 2.0 in the card's metadata, with the card's licence link pointing to Google's Gemma 4 licence page, which on 21 October 2026 redirected to Google's plain Apache 2.0 page. No LICENSE file in the repository. The card's Ethics and Safety section adds that deployments must adhere to the Gemma Prohibited Use Policy, which Google says it may update; Gemma 4's cards carry no such line |\n| Runtimes | Transformers 5.19.0 and later (embedding_gemma2) with Sentence Transformers 6.1.0 or later, vLLM 0.32.0 (EmbeddingGemma2Model in its registry), SGLang 0.5.22 (embedding_gemma2.py), llama.cpp (gemma-embedding2 architecture, with ggml-org's GGUF and mmproj files), Ollama 0.40.13 (its MLX runner's gemma4embedding model and import code). Transformers.js 4.3.2 runs the ONNX build in the browser. LM Studio's catalogue and Ollama's library weren't checked |\n| Builds | Google publishes BF16 only in its own organisation. Others publish quantised builds: onnx-community's ONNX (q4 about 0.5 GB with all three encoders), litert-community's LiteRT-LM int4 QAT builds (165 MB text, 388 MB text and vision, 485 MB full, plus NPU builds), ggml-org's BF16 and Q8_0 GGUF with a separate multimodal projector, unsloth's GGUF and mlx-community's MLX |\n| Training data | Web documents in over 140 languages, code, images, video, audio and paired cross-modal samples, with a January 2025 cut-off, filtered for CSAM, personal and sensitive data, and quality. No sources, mix or sizes |\n| Hardware | The card names mobile devices and laptops as targets and gives text-only, text and image, text and audio and full loading sizes (270M, 440M, 570M, 740M parameters) but no memory figure. The BF16 files are 1.53 GB |\n| Capabilities | embed.text, embed.multimodal, embed.multilingual, embed.code, model.open-weights |\n| Tags | open-weights, model, embeddings, multimodal, vision, audio, video, multilingual, matryoshka, apache-2.0, safetensors, sentence-transformers, self-hosted |\n| JSON | https://www.anchorterminal.com/api/v1/tools/embeddinggemma-2.json |\n\n## Score breakdown (methodology v0.7, October 2026 research run)\n\nAssessed 2026-10-30 from public evidence against the published checklist (https://www.anchorterminal.com/benchmark/#checklist). Confidence: medium. Performance and Task success pending (no score, not in the total); the total is Σ(score × weight) ÷ 80 over the 7 assessed categories. \"This run\" is each category's share of the 100 points.\n\n| Category | Weight | This run | Score (0–100) | Points |\n| --- | --- | --- | --- | --- |\n| Capability | 35% | 35 | 59 | 20.6 |\n| Reach | 25% | 25 | 75 | 18.8 |\n| Risk | 20% | 20 | 60 | 12.0 |\n| Agent interface | 10% | 10 | 70 | 7.0 |\n| Documentation and upkeep | 10% | 10 | 80 | 8.0 |\n| Negative events | up to −15 | up to −15 | none recorded | 0 |\n| **Total** | | | | **66.4 → B** |\n\n### Why each score\n\n- Capability 59 (checked 2026-10-30): From 9 benchmark results the publisher submitted (counted at half weight) in 2 areas, each scored as its share of the way from the benchmark's floor to the best result any model has on the same leaderboard. By area, Retrieval 63, Other MTEB task types 53. Publisher-submitted results count at half weight, and numbers that appear only on the publisher's card don't count.\n- Reach 75 (checked 2026-10-30): Memory earns the full 30, since Google's BF16 files are 1.53 GB and 4-bit builds by others are about 0.5 GB. Transformers, vLLM, SGLang, llama.cpp and Ollama each run it in a released version (25). Builds score 0 because every quantised build sits in another organisation. The card applies the Gemma Prohibited Use Policy on top of Apache 2.0, so terms read as a custom licence with free commercial use (10). No gate (10). Hosted too scores 0, with no inference provider on Hugging Face.\n- Risk 60 (checked 2026-10-30): File safety gets the full 25 for one safetensors file loaded by Transformers' own code. Provenance gets 15, Google's verified organisation and a checksum on every file. Safety evaluation is 0, the card has none. Agent guidance gets 10, since the card says there was no safety tuning and names the safeguards to add. Licence risk falls to 5, 5 off each for the use policy and Google's right to update it. The security programme gets 5 for Google's security.txt.\n- Agent interface 70 (checked 2026-10-30): Read for an embedding model, the template, tool-calling, trained tool calling and structured-output lines stand for the prompt format and what it takes and returns. The template and named task prefixes ship in the repository (20), every modality's input and the output are documented (25), it was trained for task prefixes with one vector per input (10), and each vector size and how to get it is documented (15). Reasoning control and context score 0, with no reasoning and an 8,192-token input.\n- Documentation and upkeep 80 (checked 2026-10-30): The card covers all seven items (20) and describes the training data by type with a cut-off (10). Google changed the repository two days ago (20). Google members answered 4 of 12 discussions (10). The earlier model keeps its own repository and gemma.ai dates both (10). Config, tokeniser, template and card agree, and no broken-file report waits (10).\n\nFix list for a coding agent, everything this grade says the listing lacks, the biggest gain first (16 items): https://www.anchorterminal.com/fixes/embeddinggemma-2.md (JSON https://www.anchorterminal.com/fixes/embeddinggemma-2.json)\n\n### What we couldn't check\n\n- Terms and licence risk read the card's line that deployments must adhere to the Gemma Prohibited Use Policy as a term added to Apache 2.0. Read as a request, terms is 15 and licence risk 15, Reach 80 and Risk 70\n- Unchecked in this session, ai.google.dev (documentation, licence and policy pages), blog.google, gemma.ai and arXiv, since the fetcher refused them; their facts come from the 21 October 2026 checks of EmbeddingGemma 300M and Gemma 4\n- If the card's Prohibited Use Policy link still shows only the moved-documentation note it showed on 21 October, a strict reading takes licence risk to 0\n- Builds would be 10 if a Google page links litert-community's int4 builds as Google's own; none read here does\n- Unchecked, LM Studio's catalogue and Ollama's library for an EmbeddingGemma 2 entry; neither changes the runtimes score\n- Unchecked today, whether the Gemini API or Vertex AI serves EmbeddingGemma 2\n- Agent guidance takes the 10 band on the card's statement that there was no safety tuning plus named safeguards; EmbeddingGemma 300M's card, with general cautions only, got 5\n\n### Sources\n\n- model card, licence link, inference providers: \u003chttps://huggingface.co/google/embeddinggemma-2\u003e (seen 2026-10-30)\n- not gated, created 14 September 2026, last modified 28 October 2026, 744,371,512 BF16 parameters, downloads: \u003chttps://huggingface.co/api/models/google/embeddinggemma-2\u003e (seen 2026-10-30)\n- file list, sizes and checksums: \u003chttps://huggingface.co/api/models/google/embeddinggemma-2/tree/main?recursive=true\u003e (seen 2026-10-30)\n- four commits, 6 to 28 October 2026: \u003chttps://huggingface.co/api/models/google/embeddinggemma-2/commits/main\u003e (seen 2026-10-30)\n- 21 discussions, 12 not pull requests, counted by author, member flag and date, text not read: \u003chttps://huggingface.co/api/models/google/embeddinggemma-2/discussions\u003e (seen 2026-10-30)\n- architecture, sliding window, max positions: \u003chttps://huggingface.co/google/embeddinggemma-2/raw/main/config.json\u003e (seen 2026-10-30)\n- task prefixes, Sentence Transformers 6.1.0 requirement: \u003chttps://huggingface.co/google/embeddinggemma-2/raw/main/config_sentence_transformers.json\u003e (seen 2026-10-30)\n- Google's EmbeddingGemma repositories, no quantised EmbeddingGemma 2 build: \u003chttps://huggingface.co/api/models?author=google\u0026search=embeddinggemma\u003e (seen 2026-10-30)\n- ONNX build for Transformers.js with WebGPU example: \u003chttps://huggingface.co/onnx-community/embeddinggemma-2-ONNX\u003e (seen 2026-10-30)\n- LiteRT-LM int4 QAT builds: \u003chttps://huggingface.co/litert-community/embeddinggemma-2-740m-litert-lm\u003e (seen 2026-10-30)\n- llama.cpp organisation's GGUF and mmproj: \u003chttps://huggingface.co/ggml-org/embeddinggemma-2-GGUF\u003e (seen 2026-10-30)\n- Transformers release with embedding_gemma2, sliding window documented as a radius: \u003chttps://raw.githubusercontent.com/huggingface/transformers/v5.19.0/src/transformers/models/embedding_gemma2/configuration_embedding_gemma2.py\u003e (seen 2026-10-30)\n- vLLM EmbeddingGemma2Model: \u003chttps://raw.githubusercontent.com/vllm-project/vllm/v0.32.0/vllm/model_executor/models/registry.py\u003e (seen 2026-10-30)\n- SGLang EmbeddingGemma2Model: \u003chttps://raw.githubusercontent.com/sgl-project/sglang/v0.5.22/python/sglang/srt/models/embedding_gemma2.py\u003e (seen 2026-10-30)\n- llama.cpp gemma-embedding2 architecture: \u003chttps://raw.githubusercontent.com/ggml-org/llama.cpp/master/src/llama-arch.cpp\u003e (seen 2026-10-30)\n- Ollama EmbeddingGemma2Model in the MLX runner (partial clone): \u003chttps://github.com/ollama/ollama/tree/v0.40.13/mlxrunner/model/gemma4embedding\u003e (seen 2026-10-30)\n- Transformers.js embedding_gemma2 (partial clone): \u003chttps://github.com/huggingface/transformers.js/tree/4.3.2/packages/transformers/src/models/embedding_gemma2\u003e (seen 2026-10-30)\n- MTEB rows for EmbeddingGemma 2 and 300M: \u003chttps://mteb-leaderboard-backend.hf.space/v1/benchmarks/MTEB%28Multilingual%2C%20v2%29/scores\u003e (seen 2026-10-30)\n- dated releases, per the 21 October 2026 check: \u003chttps://gemma.ai/models/releases.md\u003e (seen 2026-10-30)\n- Google security.txt, per the 21 October 2026 check: \u003chttps://www.google.com/.well-known/security.txt\u003e (seen 2026-10-30)\n- Independent benchmark result used for Capability: \u003chttps://huggingface.co/spaces/mteb/leaderboard\u003e (seen 2026-10-30)\n\n## Score history\n\nEvery move of a score since the listing was graded, newest first, with its cause. A score can't move without an entry here. JSON: https://www.anchorterminal.com/api/v1/tools/embeddinggemma-2/history.json\n\n- 2026-10-30: graded 66.4 B (methodology 0.7)\n\n## Who's behind it (provenance 70/100, checked 2026-10-30)\n\n| Check | Finding | Points |\n| --- | --- | --- |\n| Legal entity named | Google LLC (RDAP registrant of gemma.ai; the card names Google DeepMind as author) | 20/20 |\n| Domain age | gemma.ai, registered 2017-12-16 (8 years) | 11/15 |\n| Endpoint on the vendor's domain | no hosted endpoint | n/a |\n| Terms of service | read, states 3 of the 7 things a reader expects | 6.6/10 |\n| Privacy policy | nothing hosted, not scored | n/a |\n| Status page | not found | 0/10 |\n| Changelog | published | 10/10 |\n| security.txt | not found | 0/10 |\n| People named | 6 named with their role, 2 with a public profile | 10/10 |\n| Support route | help pages, a community | 5/5 |\n\nThe registrant, registration date, releases page and security.txt findings are carried from the embeddinggemma-300m and gemma-4 facts checked on 21 October 2026; gemma.ai and ai.google.dev couldn't be fetched in this session.\n\nNothing is hosted, so the licence stands in for terms of service and no privacy policy is scored.\n\nThe repository was created privately on 14 September 2026 and published on 6 October 2026 (the super-squash commit, MTEB's release date and Transformers' contribution note agree). The card last changed on 28 October 2026.\n\ngemma.ai/.well-known/security.txt answered 404 on 21 October 2026. Google's company-wide security.txt points to its vulnerability reward programme.\n\n### Terms and privacy, as read\n\nA reading by a fixed set of rules, each answered with the vendor's own sentence. Not legal advice.\n\n**Terms of service** (https://ai.google.dev/gemma/docs/gemma_4_license), read 2026-10-30, dated 2026-04-01, states 3 of the 7 things a reader expects.\n\n- Gives the date it was last updated. Last updated 2026-04-01.\n- Not found in the text. Names the governing law or courts.\n- Not found in the text. Says how the agreement or account can be ended.\n- Not found in the text. Says how changes to the terms are announced.\n- Not found in the text. Refers to a service level or uptime commitment.\n\n**Privacy policy**. Nothing is hosted by the vendor, so there is no privacy policy to read and the check isn't scored.\n\n\n## Humans behind the project (researched 2026-10-30 from public pages, not part of the grade)\n\n### Key people\n\n- Demis Hassabis, Co-founder and CEO, Google DeepMind\n- Shane Legg, Co-founder and Chief AGI Scientist, Google DeepMind\n- Mustafa Suleyman, Co-founder and former head of applied AI, DeepMind\n- Gus Martins, Gemma Product Manager. LinkedIn: https://uk.linkedin.com/in/lgusmartins. X: https://x.com/gusthema\n- Henrique Schechter Vera, Lead author, EmbeddingGemma 2 technical report\n- Sahil Dua, Author, EmbeddingGemma 2 technical report. X: https://x.com/sahildua2305\n- Note: the founders, gus martins and sahil dua's x account are carried from data/humans/embeddinggemma-300m.json (checked 28 october 2026), the same company. embeddinggemma 2's model card, read on 30 october 2026, cites its technical report (arxiv 2610.37364) as by henrique schechter vera, sahil dua and the embeddinggemma team; both also appear among the hugging face accounts in the repository's first commit. the launch blog, the report itself and google's documentation couldn't be fetched in this session, so no other role is listed and no account is given for henrique schechter vera. ownership is carried from the 300m file (subsidiary of alphabet inc., checked 28 october 2026); a subsidiary gets no investors.\n- Google DeepMind: LinkedIn https://www.linkedin.com/company/googledeepmind/ X https://x.com/googlegemma GitHub https://github.com/google-gemma\n\n### Ownership and investors\n\n- Ownership: Part of Alphabet Inc.\n\n### Getting support\n\n- Discord: https://discord.gg/google-dev-community\n- Help: https://huggingface.co/google/embeddinggemma-2/discussions\n- Help: https://ai.google.dev/gemma/docs/embeddinggemma\n- Community (Model discussions): https://huggingface.co/google/embeddinggemma-2/discussions\n- Sources: https://huggingface.co/google/embeddinggemma-2, https://huggingface.co/api/models/google/embeddinggemma-2/commits/main, https://arxiv.org/abs/2610.37364, https://deepmind.google/about/, https://en.wikipedia.org/wiki/Google_DeepMind, https://en.wikipedia.org/wiki/Demis_Hassabis, https://en.wikipedia.org/wiki/Shane_Legg, https://en.wikipedia.org/wiki/Mustafa_Suleyman, https://uk.linkedin.com/in/lgusmartins, https://x.com/gusthema, https://x.com/sahildua2305, https://ai.google.dev/gemma/docs/discord\n\n## Probe metrics\n\nNot measured yet. Our benchmark probes haven't run, so there's no availability, latency or error rate from a run and Performance is pending. Live uptime, where we poll the endpoint, is under Live and doesn't change the score.\n\nNo hosted version of these weights has a price on OpenRouter, so what a task costs is your own hardware and power.\n\n## Strengths\n\n- Apache 2.0, not gated, with safetensors only in Google's verified organisation and a checksum on every file\n- Text, images, video and audio in one 8,192-token input and one 768-dimension vector, truncatable to 512, 256 or 128\n- Runs in Transformers 5.19.0 with Sentence Transformers 6.1.0, vLLM 0.32.0, SGLang 0.5.22, llama.cpp and Ollama 0.40.13\n- Vision and audio encoders load only when needed, so text alone is 270M parameters\n- A card that documents every modality's token cost, the task prefixes and a January 2025 cut-off\n\n## Weaknesses\n\n- The card requires deployments to follow the Gemma Prohibited Use Policy, which Google can update\n- Google publishes no quantised build of its own; the 4-bit and GGUF builds are other organisations'\n- An 8,192-token input shared by every modality, about 29 images or 327 seconds of audio\n- No safety results, and no inference provider hosts it\n- MTEB's rows for it sit below the card's figures on every tab read, and below EmbeddingGemma 300M on the two text tabs\n\n\n\n## Before you call it (notes for agents)\n\n1. Install sentence-transformers 6.1.0 or later and transformers 5.19.0 or later; older Sentence Transformers ignore the order of multimodal dict inputs\n2. Use prompt_name SearchQuery for queries and Document for untitled documents; write `title: {title} | text: {content}` yourself for titled ones, and pass images, video and audio with no prefix\n3. For text only, load with config_kwargs {\"vision_config\": None, \"audio_config\": None} to skip 470M parameters of encoders\n4. Run in bfloat16 or float32, never float16, which returns NaN or degraded vectors without an error\n5. Pass truncate_dim and normalize_embeddings=True for smaller vectors, and keep queries and documents at the same size\n6. Supply audio as mono 16 kHz; video is sampled at 1 frame a second by default\n\n## Get started\n\nInstall:\n\n```bash\npip install -U \"sentence-transformers\u003e=6.1.0\" \"transformers\u003e=5.19.0\"\npython -c \"from sentence_transformers import SentenceTransformer; m=SentenceTransformer('google/embeddinggemma-2', config_kwargs={'vision_config': None, 'audio_config': None}); print(m.encode('Which planet is red?', prompt_name='SearchQuery').shape)\"\n```\n\n## Similar tools\n\nRanked by shared capabilities, then score. Same-category tools with no shared capability key are listed last.\n\n| Tool | Grade | Score | Rank | Shared capabilities | x402 | Markdown |\n| --- | --- | --- | --- | --- | --- | --- |\n| Qwen3-Embedding-8B | BB | 74.7 | open-weight #1 | embed.text, embed.multilingual, embed.code, model.open-weights | no | https://www.anchorterminal.com/tools/qwen3-embedding-8b.md |\n| Qwen3-Embedding-4B | BB | 72.9 | open-weight #3 | embed.text, embed.multilingual, embed.code, model.open-weights | no | https://www.anchorterminal.com/tools/qwen3-embedding-4b.md |\n| Gemini Embedding | BB | 70.6 | 170 | embed.text, embed.multimodal, embed.code, embed.multilingual | no | https://www.anchorterminal.com/tools/gemini-embedding.md |\n| Qwen3-Embedding-0.6B | B | 63.5 | open-weight #25 | embed.text, embed.multilingual, embed.code, model.open-weights | no | https://www.anchorterminal.com/tools/qwen3-embedding-0-6b.md |\n| Qwen3-VL-Embedding-8B | B | 62.1 | open-weight #32 | embed.text, embed.multimodal, embed.multilingual, model.open-weights | no | https://www.anchorterminal.com/tools/qwen3-vl-embedding-8b.md |\n| Jina Embeddings and Reranker | C | 61.2 | 559 | embed.text, embed.multimodal, embed.code, embed.multilingual | no | https://www.anchorterminal.com/tools/jina-embeddings.md |\n\n## Panel reviews (0)\n\nReviewed by the Anchor panel (https://www.anchorterminal.com/reviewers/index.md): .\n\nDesk reviews, written from public documentation, pricing, terms, source and status history on 1 October 2026. No calls made. For a desk review, the outcome says whether the reviewer's questions could be answered from public material: success, partial or failure. How reviews work: https://www.anchorterminal.com/reviews/how-it-works.md\n\n## Notable\n\n- 338,269 downloads in the last 30 days and 3,026 likes on Hugging Face on 30 October 2026 (source: \u003chttps://huggingface.co/api/models/google/embeddinggemma-2\u003e)\n- The MTEB leaderboard lists it with every task scored on MTEB(Multilingual, v2) at 60.52 mean task score, MTEB(eng, v2) at 67.53, MTEB(Code, v1) at 77.88, MIEB(lite) at 63.52 and MAEB(v1) at 49.31. Each is below the figure on Google's card (61.36, 68.46, 78.68, 65.67 and 49.87), and on the two text tabs below EmbeddingGemma 300M's own rows (61.14 and 69.67). MTEB doesn't say who ran them (source: \u003chttps://mteb-leaderboard-backend.hf.space/v1/benchmarks/MTEB%28Multilingual%2C%20v2%29/scores\u003e)\n- Google's card claims about a 14 per cent improvement on code tasks over EmbeddingGemma 300M. The MTEB(Code, v1) rows give 77.88 against 68.76, 13 per cent (source: \u003chttps://huggingface.co/google/embeddinggemma-2\u003e)\n- On 27 October 2026 Google raised the card's MIEB(lite) figure from 64.64 to 65.67 and its MAEB figure from 49.39 to 49.87, with no note on the card (source: \u003chttps://huggingface.co/api/models/google/embeddinggemma-2/commits/main\u003e)\n- The browser claim from the launch has a published build. onnx-community/embeddinggemma-2-ONNX holds fp32, fp16, 8-bit, q4 and q4f16 ONNX files for Transformers.js with a WebGPU example, and Transformers.js 4.3.2 ships an embedding_gemma2 model. Hugging Face's ONNX community published it, not Google, and we haven't run it (source: \u003chttps://huggingface.co/onnx-community/embeddinggemma-2-ONNX\u003e)\n- EmbeddingGemma 300M, the earlier text-only model under the Gemma Terms of Use, stays published in its own repository and is listed separately\n\n- Best embedding and reranking APIs for AI agents: https://www.anchorterminal.com/best/embeddings/index.md\n- All 81 embeddings comparisons: https://www.anchorterminal.com/compare/embeddings/index.md\n\n## Compare\n\n- [EmbeddingGemma 2 vs EmbeddingGemma 300M](https://www.anchorterminal.com/compare/embeddinggemma-2-vs-embeddinggemma-300m.md): B 66.4 vs D 50.7\n- [EmbeddingGemma 2 vs Granite Embedding Small English R2](https://www.anchorterminal.com/compare/embeddinggemma-2-vs-granite-embedding-small-english-r2.md): B 66.4 vs E 41.7\n- [EmbeddingGemma 2 vs jina-embeddings-v5-omni-small](https://www.anchorterminal.com/compare/embeddinggemma-2-vs-jina-embeddings-v5-omni-small.md): B 66.4 vs D 50.2\n- [EmbeddingGemma 2 vs Qwen3-Embedding-0.6B](https://www.anchorterminal.com/compare/embeddinggemma-2-vs-qwen3-embedding-0-6b.md): B 66.4 vs B 63.5\n- [EmbeddingGemma 2 vs Qwen3-Embedding-4B](https://www.anchorterminal.com/compare/embeddinggemma-2-vs-qwen3-embedding-4b.md): B 66.4 vs BB 72.9\n- [EmbeddingGemma 2 vs Qwen3-Embedding-8B](https://www.anchorterminal.com/compare/embeddinggemma-2-vs-qwen3-embedding-8b.md): B 66.4 vs BB 74.7\n- [EmbeddingGemma 2 vs Qwen3-VL-Embedding-2B](https://www.anchorterminal.com/compare/embeddinggemma-2-vs-qwen3-vl-embedding-2b.md): B 66.4 vs C 55.4\n- [EmbeddingGemma 2 vs Qwen3-VL-Embedding-8B](https://www.anchorterminal.com/compare/embeddinggemma-2-vs-qwen3-vl-embedding-8b.md): B 66.4 vs B 62.1\n\n## Verify this listing\n\nFor the vendor. The badge or a plain link to this page verifies the listing, from a page on gemma.ai or one of its subdomains. It shows the listing is the vendor's and that the vendor knows it's here, and it never changes a grade, rank or review. The vendor sends the page's address to `POST https://www.anchorterminal.com/api/v1/verify` as `{\"slug\": \"embeddinggemma-2\", \"url\": \"…\"}`, or calls the `verify_listing` tool at https://www.anchorterminal.com/mcp. We fetch the page once, then again every week; two failed checks in a row and the verification lapses, and a later pass restores it. What we check: https://www.anchorterminal.com/builders/index.md#verify\n\nHTML badge:\n\n```html\n\u003ca href=\"https://www.anchorterminal.com/tools/embeddinggemma-2\"\u003e\u003cimg src=\"https://www.anchorterminal.com/badges/embeddinggemma-2.svg\" alt=\"EmbeddingGemma 2 on Anchor Terminal\" height=\"20\"\u003e\u003c/a\u003e\n```\n\nMarkdown badge, for a README:\n\n```markdown\n[![EmbeddingGemma 2 on Anchor Terminal](https://www.anchorterminal.com/badges/embeddinggemma-2.svg)](https://www.anchorterminal.com/tools/embeddinggemma-2)\n```\n\nPlain link:\n\n```html\n\u003ca href=\"https://www.anchorterminal.com/tools/embeddinggemma-2\"\u003eEmbeddingGemma 2 on Anchor Terminal\u003c/a\u003e\n```\n\n## Share this listing\n\nFor the vendor. Sharing assets for social media, two PNGs of 1200 × 630 that say EmbeddingGemma 2 is listed on Anchor Terminal, with the vendor's logo and this page's address and no grade or score.\n\n- Dark: https://www.anchorterminal.com/assets/share/embeddinggemma-2-dark.png\n- Light: https://www.anchorterminal.com/assets/share/embeddinggemma-2-light.png\n- Create a video and download its frames: https://www.anchorterminal.com/assets/listing-video/index.html?listing=embeddinggemma-2\n", "meta": { "attribution": "Anchor Terminal (https://www.anchorterminal.com)", "attributionRequired": true, "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)", "docs": "https://www.anchorterminal.com/docs/", "generatedAt": "2026-10-31", "license": "CC-BY-4.0", "method": "https://www.anchorterminal.com/benchmark/", "methodology": "0.7", "openapi": "https://www.anchorterminal.com/openapi.json", "preview": false, "run": "2026-10-01", "runLabel": "October 2026 research run", "terms": "https://www.anchorterminal.com/terms/#data" }, "page": { "breadcrumbs": [ { "name": "Home", "url": "https://www.anchorterminal.com/" }, { "name": "Terminal", "url": "https://www.anchorterminal.com/tools/" }, { "name": "Embeddings \u0026 rerankers", "url": "https://www.anchorterminal.com/categories/embeddings" }, { "name": "EmbeddingGemma 2", "url": "" } ], "description": "EmbeddingGemma 2 is Google DeepMind's open-weight multimodal embedding model of 740M parameters, mapping text, code, images, video and audio into one 768-dimension vector space from up to 8,192 tokens, under Apache 2.0.", "facts": [ "#7 in Embeddings", "None auth", "0 desk reviews" ], "h1": "EmbeddingGemma 2", "image": "https://www.anchorterminal.com/assets/og/tools-embeddinggemma-2.png", "path": "/tools/embeddinggemma-2", "published": "2026-10-01", "section": "tools", "title": "EmbeddingGemma 2 review (2026): pricing, alternatives and grade B", "toc": null, "updated": "2026-10-30", "url": "https://www.anchorterminal.com/tools/embeddinggemma-2" }, "tokens": { "markdown": 7550, "slim": 2180 }, "version": 1 }