Embedding

MITOpen weightsUpdated August 2026

e5-mistral-7b-instruct

Instruction-tuned E5 embedding model built on a Mistral 7B backbone for text embedding and retrieval tasks.

Microsoft / intfloat · E5

Editorial review

Reviewed byOpenSourcesAI EditorialLast updatedAugust 2026SourcesExact model card

Model checkpoints, context windows, provider support, local runtime compatibility, and license terms can change quickly. Verify the exact model card before production or commercial use.

Best for

Teams comparing larger instruction-tuned embedding models for retrieval quality when smaller E5 checkpoints are not enough.

Who should use it

  • Teams comparing larger instruction-tuned embedding models for retrieval quality when smaller E5 checkpoints are not enough.
  • Builders who want local or self-hosted testing options.
  • RAG builders comparing retrieval quality, latency, and multilingual coverage.

Common workflows

  • Instruction-tuned embedding workflows
  • embedding workflows
  • rag workflows
  • semantic search workflows
  • retrieval workflows

Deployment and hardware notes

~14.2 GB in fp16; a 16 GB GPU serves it, and quantized variants shrink that at some quality cost you should measure on your corpus.

License and usage notes

MIT. Open weights. Verify the exact model card and license terms for the checkpoint or hosted provider you use.

Strengths

  • Open weights model option for E5 workflows.
  • Teams comparing larger instruction-tuned embedding models for retrieval quality when smaller E5 checkpoints are not enough.
  • Task instructions go on the query side only ('Instruct: {task} Query: {q}'); documents are embedded without instructions. Embeddings come from last-token pooling, not CLS.

Limitations

  • Each embedding is a 7.1B-parameter forward pass — interactive query latency effectively requires a GPU, and 4096-dimension vectors quadruple index storage against 1024-dimension models.
  • ~14.2 GB in fp16; a 16 GB GPU serves it, and quantized variants shrink that at some quality cost you should measure on your corpus.
  • Context window and limits: 32,768 tokens.
  • Verify the exact model card, provider docs, license, and serving support before production use.

Local workflow notes

Task instructions go on the query side only ('Instruct: {task} Query: {q}'); documents are embedded without instructions. Embeddings come from last-token pooling, not CLS.

Local runtimes: Ollama where supported, LM Studio where supported, llama.cpp where supported, Transformers

Platforms: Windows, macOS, Linux

Embedding spec

Memory~14.2 GB in fp16 (7.1B parameters) — an LLM-scale embedding model · ~4.3 GB at Q4_K_MOutput4096-dimension vectorsMax input32,768 tokens

A Mistral-7B turned embedder: 4096-dimension vectors and a 32,768-token ceiling, though the card's own examples stay near 4k. English-focused despite the multilingual base.

Sources to verify

Exact model card

Related resources

Continue with model source notes, local tools, and implementation guides related to this model.

E5 family hubEmbedding models shortlistCompare BGE vs E5RAG for documents stackAI tools directory
Hardware~14.2 GB in fp16 (7.1B parameters) — an LLM-scale embedding modelRuntimeOllama or LM Studio where supported, llama.cpp, Transformers, vLLMContext32,768 tokensLast updated2026
Exact model card →

Model ecosystem connections

Use these next-step links to move from this profile into related tools, comparisons, guides, stacks, and curated shortlists.

Recommended runtimes and tools

Ollama →LM Studio →llama.cpp →Transformers →vLLM →

Setup and deployment

Check if this model fits your GPU →What can 16GB VRAM run? →What is quantization? GGUF, Q4, Q8 explained →What is VRAM and how much do you need? →

Related model pages

E5 family hub →e5-small →e5-small-v2 →Multilingual E5 Large →bge-base-en-v1.5 →bge-embedding-gemma2 →bge-large-en-v1.5 →

Guides, stacks, and comparisons

Embedding models shortlist →Compare BGE vs E5 →RAG for documents stack →All model comparisons →

Related models

e5-smallLegacy lightweight embedding baselinee5-small-v2Local or cost-sensitive embedding pipelines that need a practical English retrieval model without the overhead of larger E5 releases.Multilingual E5 LargeTeams building multilingual retrieval, semantic search, and RAG pipelines.bge-base-en-v1.5Teams building English search or RAG systems on modest GPUs without dropping to the smallest embedding tier.bge-embedding-gemma2Gemma-based embedding experimentsbge-large-en-v1.5English retrieval stacks that want a stronger embedding baseline for search and RAG before moving to multilingual models.

Ready to run this model locally?

Find a compatible interface in our Local AI Tools directory →