BGE-M3 has no published price in our catalogue yet. Prices appear here when a host lists it. It returns 1,024-dimension vectors and takes up to 8,192 tokens per input; its record lists text input.Not on the atlas: the atlas maps text-model prices, and embedding models are billed on input only. Compare embedding models by price, dimensions and max input instead.
Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
Cite
Cite this page
ModelsAtlas. “BGE-M3”. Retrieved 1 Nov 2026. https://modelsatlas.com/models/baai/bge-m3
- Input / 1M
- —
- Dimensions
1,024
- Context
- tokens
8K
- Offers
- 0 offers
No standard-tier price
Default vector size
≈ 6,000 words
No host lists it yet
Hosts and pricing
0 offers · tiers are rows, not pagesSuggest a correction
No host in our catalogue lists BGE-M3 yet, so there is no price to show. The specs below are what we hold about the model itself.
Specifications
Tap src for a sourceSuggest a correction
| Catalogue id | baai-bge-m3 | |
|---|---|---|
| Creator | BAAI | |
| Context window | 8,192 tokens | Asserted |
| Embedding dimensions | 1,024 | Asserted |
| Modalities | text | Asserted |
| Licence | mit | Asserted |
| Weights | Open weights · Hugging Face ↗ | Asserted |
| Variants | One (base) | |
| Record checked | 2026-11-01 |
Capabilities
“Not disclosed” means our records do not say, not that the model lacks it.
- Vision inputNot listed
- AudioNot listed
- Open weightsPublished
- Free tierNone listed
- Batch tierNone listed
We hold no data yet on tool calling, structured output or benchmark scores, so none are shown.
Alternatives to BGE-M3
Embedding models are compared with each other, not on the atlas: see all embedding models by price, dimensions and max input.
Works with
Where BGE-M3 connects across the atlas
◉Tools built on it
- None linked yet. Tell us about one
Questions about BGE-M3
How much does BGE-M3 cost?
No host in our catalogue lists a price for it. Its weights are open, so you can run it on your own hardware.
How many dimensions does BGE-M3 output?
1,024 by default, per its model card. It takes up to 8,192 tokens per input.
What is the BGE-M3 context window?
8,192 tokens, roughly 6,000 English words in one request.
What modalities does BGE-M3 support?
Its record lists text. We do not hold capability data beyond that, so tool calling and structured output are not stated here.
Is there a free tier for BGE-M3?
No free-tier offer is listed for BGE-M3 in our catalogue.
Can I run BGE-M3 myself?
Its record marks it as open-weight, under the mit licence, and links the weights on Hugging Face. Check the licence terms before using it commercially.
Why does BGE-M3 have no grid square?
The atlas places a model by its standard-tier input price and its context window. We hold a context window for it but no standard-tier price, so it is not on the map yet.
Sources
Every number on this page, where it came from
- huggingface.co/BAAI/bge-m3/blob/main/README.md — context window, modalities, licenceAsserted · 1 Nov 2026
- Our calculation — grid reference, monthly cost and percentage comparisonsDerived · 1 Nov 2026
Spotted something wrong? Report a correction · Methodology
Keep going in your AI
Opens with this page as the source. We add nothing else to the prompt.
- How much does BGE-M3 cost?
- How many dimensions does BGE-M3 output?
- What is the BGE-M3 context window?