Gemini Embedding
GoogleGoogle’s embedding line on the Gemini API and Vertex AI. The head model is gemini-embedding-2, Google’s first natively multimodal embedder, which maps text, images, video, audio and PDFs into one space; gemini-embedding-001, the text-only predecessor with 100-plus languages, Matryoshka dimensions down from 3,072 and a 2,048-token input, remains available.
The version the brief expected has moved: gemini-embedding-001 is no longer the head of the line. Google’s embeddings docs call gemini-embedding-2 “the latest model” and both carry list prices, with -001 kept for text-only use. The 68.32 MTEB multilingual figure belongs to -001; the successor reports 69.9 on the same instrument. Neither is downloadable. Verified 2026-09-11 via the Gemini API embeddings docs, the Gemini API pricing page, the Gemini Embedding 2 technical report and the 2025 GA announcement.
Openness
1 high confidence- weights
- closed(no checkpoint distributed for gemini-embedding-2 or gemini-embedding-001
- data
- closed(training corpus not released
- code
- closed(no training or inference implementation
- license
- proprietary(hosted API under Google’s terms
API-only across both live SKUs, so the ladder's first rule fires. The published technical report is the one thing that distinguishes this from the rest of the cluster — it makes the numbers auditable — but a paper is not a corpus, weights or a pipeline, and the ladder scores what ships.
- https://ai.google.dev/gemini-api/docs/embeddings recorded 2026-09-11
'The latest model, gemini-embedding-2, is the first multimodal' embedding model, and 'for text-only use cases, gemini-embedding-001 remains available'. Both are invoked as embedContent calls against generativelanguage.googleapis.com; no download is offered.
- https://ai.google.dev/gemini-api/docs/pricing recorded 2026-09-11
The Gemini API pricing page lists Gemini Embedding 2 (gemini-embedding-2) and Gemini Embedding (gemini-embedding-001) as priced hosted models, which is the only way either is sold.
- https://arxiv.org/abs/2605.27295 recorded 2026-09-11
The Gemini Embedding 2 report describes 'large-scale contrastive learning in a multi-task multi-stage training setup' and releases no corpus, checkpoint or training code.
Adoption
4 medium confidenceNo artifact to count and no call volume published. The level rests on the distribution surface: general availability across both the Gemini API and Vertex AI, a free tier in Google AI Studio alongside paid limits, and a migration path Google enforced by deprecating embedding-001 and text-embedding-004 into this line. Directional, so confidence stops at medium.
- https://developers.googleblog.com/en/gemini-embedding-available-gemini-api/ recorded 2026-09-11
gemini-embedding-001 'is now generally available to developers in the Gemini API and Vertex AI'; free and paid tiers are offered, and the legacy embedding-001 and text-embedding-004 models were deprecated on 2025-08-14 and 2026-01-14 respectively, migrating their users onto this line.
- https://ai.google.dev/gemini-api/docs/pricing recorded 2026-09-11
Both gemini-embedding-2 and gemini-embedding-001 carry list prices on the Gemini API pricing page, so both are generally sold rather than previewed.
Capability
4 high confidenceRung 4, competitive frontier. The numbers are real, published and multi-instrument, and the multimodal span is wider than any peer's. It stops short of 5 on two arithmetic facts rather than a judgment: 69.9 on MTEB multilingual is below the category anchor's 74.3 on MMTEB v2, and Voyage's independent-of-Google RTEB run puts voyage-4-large 3.87% above Gemini Embedding 001. The 68.32 figure the roster carried is -001's and is now the older of the two.
- https://arxiv.org/abs/2605.27295 recorded 2026-09-11
Abstract: 'a score of 62.9 R@1 on MSCOCO, 68.8 NDCG@10 on Vatex, 69.9 on MTEB multilingual and 84.0 on MTEB Code ... surpassing the performance of specialized models'.
- https://developers.googleblog.com/en/gemini-embedding-available-gemini-api/ recorded 2026-09-11
gemini-embedding-001 'has consistently held a top spot on the Massive Text Embedding Benchmark (MTEB) Multilingual leaderboard'; supports over 100 languages with a 2,048-token maximum input and 3,072 default output dimensions.
- https://blog.voyageai.com/2026/01/15/voyage-4/ recorded 2026-09-11
Over all 29 RTEB datasets, voyage-4-large surpasses Gemini Embedding 001 by an average of 3.87%.
Verified 2026-09-11