⚡️ FlashAttention-4: up to 1.3× faster than cuDNN on NVIDIA Blackwell →

Introducing Together AI's new look →

🔎 ATLAS: runtime-learning accelerators delivering up to 4x faster LLM inference →

⚡ Together GPU Clusters: self-service NVIDIA GPUs, now generally available →

📦 Batch Inference API: Process billions of tokens at 50% lower cost for most models →

🪛 Fine-Tuning Platform Upgrades: Larger Models, Longer Contexts →

Models / Together AI
Embeddings

M2-BERT 80M 32K Retrieval

80M checkpoint of M2-BERT, pretrained with sequence length 32768, and it has been fine-tuned for long-context retrieval.

Read our Docs
read docs

About model

M2-BERT 80M 32K Retrieval generates embeddings for long-context retrieval tasks, leveraging its 80M parameters and 32K sequence length. It is suitable for applications requiring efficient information retrieval. Developed by Jon Saad-Falcon, Dan Fu, and Simran Arora.

To use this embeddings model, please follow the instructions from our Docs.

    Related models
    Embeddings
    GTE ModernBERT base
    • Input/1M
      $0.08
    • Parameters
      149.0M
    No items found.
    Embeddings
    Embeddings
    BGE-Base-EN v1.5
    • Input/1M
      $0.01
    • Parameters
      109M
    No items found.
    Embeddings
    Embeddings
    UAE-Large v1
    • Parameters
      326M
    No items found.
    Embeddings
    Model specifications
    Model data
    • Model provider
      Together AI
    • Type
      Embeddings
    • Main use cases
      Embeddings
    • Deployment
      Monthly Reserved
    • Parameters
      80M
    • Context length
      32768
    • Input price

      $0.01 / 1M tokens

    • Input modalities
      Text
    • Output modalities
      Structured Data
    • Released
      November 3, 2023
    • Last updated
      February 5, 2026
    • External link
      Provider docs
    • Category
      Embeddings
    Quickstart docs
    Deploy model