Roadmap
Overview
The following is an overview of features planned for Weaviate. By clicking the link, you can upvote the feature or engage in a discussion about it. You can also join our Slack channel to discuss the roadmap in more detail.
info
The current version of Weaviate is v1.17.2. You can check the version you're currently running at the meta endpoint.
Upvote issues of interest
We use your feedback and votes on Github pages to plan future releases.
If you find any of the issues of interest.
Follow it's link and upvote it on Github by clicking the 👍 emoji.
Please feel free to engage with us about the roadmap on Weaviate's Github or on Slack.
Planned for version 1.18
- Add where filters to BM25(F) and Hybrid Search
- Scroll through every object in a class using an id cursor
- Native Roaring Bitmap support in LSM Store for faster large-scale filters
- Improve BM25 performance in large scale setups
- Replication: Tunable Write Consistency Level
- Replication: Read-Repairs
- Make CORS configurable: Allow arbitrary headers
Planned for version 1.19
- Reindex API (Inverted Index)
- Feature Request: Stemming on Inverted Index text props
- Dynamic Scaling: Move shards around nodes, drain node, etc.
- Replication: Async repairs
- Replication: Improve dynamic scaling of replication factor
Backlog
- [Feature Request]: adding the Class
alias
functionality - FR: Support nested objects / embedding objects
- Feature request: Authentication scheme for systems long-term credentials
- Feature Request: Multi-DC (Active/Active) replication
- Support for batch patch (partial update)
- Add an
In
Filter Operator - FR: Support arbitrary amount of vector indexes on object
- Support for batch import of cross referenced objects
- Support OIDC groups in AdminList
- [text2vec-openai] support arbitrary model names
- Return vectorized query with results
- (feature request) Multiple filters on cross references should match on a single object
- Proposal: Cross-Encoders as reranking
- Proposal: autoCut in Weaviate Vector Search
- Filtering returned cross references
- Add support for wildcards for include/exclude backups
- Reindex API (Vector Index)
- (feature request) Add support for PDFs and other file formats
- (feature request) Add multi2vec-huggingface module to support CLIP models with inference API
- Feature Request: BM25/Hybrid: Support ANDing query terms (currently always ORed).
- Ability to reindex a property (column) or class (table) with a simple command
- Improve Read-only message by adding reason
- Add integration tests for OpenAI
- Unexpected tokenisation for string datatype
- Replication and sharding configuration at system level
- QnA module: Provide vector to
ask
filter clause so QnA can be used without a vectorizer - OpenAI add text-davinci-003 model
- FR: Have a graphql query that works the same way as the GET /v1/objects endpoint
- Feature Request: Partial Schema Config Update using PATCH
More Resources
If you can't find the answer to your question here, please look at the:
- Frequently Asked Questions. Or,
- Knowledge base of old issues. Or,
- For questions: Stackoverflow. Or,
- For issues: Github. Or,
- Ask your question in the Slack channel: Slack.