Skip to main content

Spice v1.6.1 (Sep 1, 2025)

· 3 min read
Jack Eadie
Token Plumber at Spice AI

Announcing the release of Spice v1.6.1! ⚡

Spice 1.6.1 is a patch release that provides improved Kafka type inference and JSON flattening support, alongside several bug fixes.

What's New in v1.6.1

Improved Kafka Type Inference: Improve Kafka type inference by configuring the number of Kafka messages sampled during schema inference. Increasing the sample size can improve the robustness and reliability of inferred schemas, especially in cases where data contains optional fields or varying structures.

Example spicepod.yml:

dataset:
- from: kafka:orders_events
name: orders
params:
schema_infer_max_records: 100 # Default 1.

For details, see the Kafka Data Connector Documentation.

Improved Kafka JSON Support: Enable nested JSON Kafka messages to be represented in flattened JSON format for the dataset schema.

Example spicepod.yml:

dataset:
- from: kafka:orders_events
name: orders
params:
flatten_json: true # default false

For example, the object:

{
"order_id": "a1f2c3d4-1111-2222-3333-444455556666",
"customer": {
"id": 101,
"name": "Alice",
"premium": true,
"contact": {
"email": "[email protected]",
"phone": "555-1234"
}
},
"discount": 5.0,
"shipped": false
}

With flatten_json: true the result is:

+------------------------+-----------+-------------+
| column_name | data_type | is_nullable |
+------------------------+-----------+-------------+
| order_id | Utf8 | YES |
| customer.id | Int64 | YES |
| customer.name | Utf8 | YES |
| customer.premium | Boolean | YES |
| customer.contact.email | Utf8 | YES |
| customer.contact.phone | Utf8 | YES |
| discount | Float64 | YES |
| shipped | Boolean | YES |
+------------------------+-----------+-------------+

With flatten_json: false or ommitted the result is:

+-------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------+
| column_name | data_type | is_nullable |
+-------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------+
| order_id | Utf8 | YES |
| customer | Struct([Field { name: "id", data_type: Int64, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "name", data_type: Utf8, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "premium", data_type: Boolean, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "contact", data_type: Struct([Field { name: "email", data_type: Utf8, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "phone", data_type: Utf8, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }]) | YES |
| discount | Float64 | YES |
| shipped | Boolean | YES |
+-------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------+

For details, see the Kafka Data Connector Documentation.

Contributors

Breaking Changes

No breaking changes.

Cookbook Updates

No new cookbook recipes added in this release.

The Spice Cookbook includes 77 recipes to help you get started with Spice quickly and easily.

Upgrading

To upgrade to v1.6.1, use one of the following methods:

CLI:

spice upgrade

Homebrew:

brew upgrade spiceai/spiceai/spice

Docker:

Pull the spiceai/spiceai:1.6.1 image:

docker pull spiceai/spiceai:1.6.1

For available tags, see DockerHub.

Helm:

helm repo update
helm upgrade spiceai spiceai/spiceai

AWS Marketplace:

🎉 Spice is now available in the AWS Marketplace!

What's Changed

Changelog

  • Fix metadata field issue by @Advayp in #6957
  • Update datafusion and datafusion-table-providers crates (#6985) by @Jeadie in #6985
  • Add flatten_json param support for Kafka connector (#6976) by @sgrebnov in #6976
  • Add schema_inference_sample_count param support for Kafka connector (#6969) by @sgrebnov in #6969
  • Add integration test for Kafka connector (#6965) by @sgrebnov in #6965
  • Skip dataset health check for IcebergTableProvider datasets by @phillipleblanc in #6995

Spice v1.6.0 (Aug 26, 2025)

· 22 min read
Sergei Grebnov
Senior Software Engineer at Spice AI

Announcing the release of Spice v1.6.0! 🔥

Spice 1.6.0 upgrades DataFusion to v48, reducing expressions memory footprint by ~50% for faster planning and lower memory usage, eliminating unnecessary projections in queries, optimizing string functions like ascii and character_length for up to 3x speedup, and accelerating unbounded aggregate window functions by 5.6x. The release adds Kafka and MongoDB connectors for real-time streaming and NoSQL data acceleration, supports OpenAI Responses API for advanced model interactions including OpenAI-hosted tools like web_search and code_interpreter, improves the OpenAI Embeddings Connector with usage tier configuration for higher throughput via increased concurrent requests, introduces Model2Vec embeddings for ultra-low-latency encoding, and improves the Amazon S3 Vectors engine to support multi-column primary keys.

What's New in v1.6.0

DataFusion v48 Highlights

Spice.ai is built on the DataFusion query engine. The v48 release brings:

Performance & Size Improvements 🚀: Expressions memory footprint was reduced by ~50% resulting in faster planning and lower memory usage, with planning times improved by 10-20%. There are now fewer unnecessary projections in queries. The string functions, ascii and character_length were optimized for improved performance, with character_length achieving up to 3x speedup. Queries with unbounded aggregate window functions have improved performance by 5.6 times via avoided unnecessary computation for constant results across partitions. The Expr struct size was reduced from 272 to 144 bytes.

New Features & Enhancements ✨: Support was added for ORDER BY ALL for easy ordering of all columns in a query.

See the Apache DataFusion 48.0.0 Blog for details.

Runtime Highlights

Amazon S3 Vectors Multi-Column Primary Keys: The Amazon S3 Vectors engine now supports datasets with multi-column primary keys. This enables vector indexes for datasets where more than one column forms the primary key, such as those splitting documents into chunks for retrieval contexts. For multi-column keys, Spice serializes the keys using arrow-json format, storing them as single string keys in the vector index.

Model2Vec Embeddings: Spice now supports model2vec static embeddings with a new model2vec embeddings provider, for sentence transformers up to 500x faster and 15x smaller, enabling scenarios requiring low latency and high-throughput encoding.

embeddings:
- from: model2vec:minishlab/potion-base-8M # HuggingFace model
name: potion
- from: model2vec:path/to/my/local/model # local model
name: local

Learn more in the Model2Dev Embeddings documentation.

Kafka Data Connector: Use from: kafka:<topic> to ingest data directly from Kafka topics for integration with existing Kafka-based event streaming infrastructure, providing real-time data acceleration and query without additional middleware.

Example Spicepod.yml:

- from: kafka:orders_events
name: orders
acceleration:
enabled: true
refresh_mode: append
params:
kafka_bootstrap_servers: server:9092

Learn more in the Kafka Data Connector documentation.

MongoDB Data Connector: Use from: mongodb:<dataset> to access and accelerate data stored in MongoDB, deployed on-premises or in the cloud.

Example spicepod.yml:

datasets:
- from: mongodb:my_dataset
name: my_dataset
params:
mongodb_host: localhost
mongodb_db: my_database
mongodb_user: my_user
mongodb_pass: password

Learn more in the MongoDB Data Connector documentation.

OpenAI Responses API Support: The OpenAI Responses API (/v1/responses) is now supported, which is OpenAI's most advanced interface for generating model responses.

To enable the /v1/responses HTTP endpoint, set the responses_api parameter to enabled:

Example spicepod.yml:

models:
- name: openai_model_using_responses_api
from: openai:gpt-4.1
params:
openai_api_key: ${ secrets:OPENAI_API_KEY }
responses_api: enabled # Enable the /v1/responses endpoint for this model

Example curl request:

curl http://localhost:8090/v1/responses \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-4.1",
"input": "Tell me a three sentence bedtime story about Spice AI."
}'

To use responses in spice chat, use the --responses flag.

Example:

spice chat --responses # Use the `/v1/responses` endpoint for all completions instead of `/v1/chat/completions`

Use OpenAI-hosted tools supported by Open AI's Responses API by specifying the openai_responses_tools parameter:

Example spicepod.yml:

models:
- name: test
from: openai:gpt-4.1
params:
openai_api_key: ${ secrets:SPICE_OPENAI_API_KEY }
tools: sql, list_datasets
responses_api: enabled
openai_responses_tools: web_search, code_interpreter # 'code_interpreter' or 'web_search'

These OpenAI-specific tools are only available from the /v1/responses endpoint. Any other tools specified via the tools parameter are available from both the /v1/chat/completions and /v1/responses endpoints.

Learn more in the OpenAI Model Provider documentation.

OpenAI Embeddings & Models Connectors Usage Tier: The OpenAI Embeddings and Models Connectors now supports specifying account usage tier for embeddings and model requests, improving the performance of generating text embeddings or calling models during dataset load and search by increasing concurrent requests.

Example spicepod.yml:

embeddings:
- from: openai:text-embedding-3-small
name: openai_embed
params:
openai_usage_tier: tier1

By setting the usage tier to the matching usage tier for your OpenAI account, the Embeddings and Models Connector will increase the maximum number of concurrent requests to match the specified tier.

Learn more in the OpenAI Model Provider documentation.

Contributors

New Contributors

Breaking Changes

No breaking changes.

Cookbook Updates

The Spice Cookbook includes 77 recipes to help you get started with Spice quickly and easily.

Upgrading

To upgrade to v1.6.0, use one of the following methods:

CLI:

spice upgrade

Homebrew:

brew upgrade spiceai/spiceai/spice

Docker:

Pull the spiceai/spiceai:1.6.0 image:

docker pull spiceai/spiceai:1.6.0

For available tags, see DockerHub.

Helm:

helm repo update
helm upgrade spiceai spiceai/spiceai

AWS Marketplace:

🎉 Spice is also now available in the AWS Marketplace!

What's Changed

Dependencies

Changelog

  • Support Streaming with Tool Calls (#6941) by @Advayp in #6941
  • Fix parameterized query planning in DataFusion (#6942) by @Jeadie in #6942
  • Update the UnableToLoadCredentials error with a pointer to docs (#6937) by @phillipleblanc in #6937
  • Fix spicecloud benchmark (#6935) by @krinart in #6935
  • [Debezium] Support for VariableScaleDecimal (#6934) by @krinart in #6934
  • Update to DF 48 (#6665) by @mach-kernel and @kczimm in #6665
  • Mark append-stream and CDC datasets as ready after first message (#6914) by @sgrebnov in #6914
  • Model2Vec embedding model support (#6846) by @mach-kernel in #6846
  • Update snapshot for S3 vector search test (#6920) by @Jeadie in #6920
  • remove [] from queryset in spicepod path for CI (#6919) by @Jeadie in #6919
  • Remove verbose tracing (#6915) by @Jeadie in #6915
  • Refactor how models supporting the Responses API are loaded (#6912) by @Advayp in #6912
  • Write tests for truncate formatting in arrow_tools and fix bug. (#6900) by @Jeadie in #6900
  • Support using the Responses API from spice chat (#6894) by @Advayp in #6894
  • Include GPT-5 into Text-To-SQL and Financebench benchmarks (#6907) by @sgrebnov in #6907
  • Better error message when credentials aren't loaded for S3 Vectors (#6910) by @phillipleblanc in #6910
  • Add tracing and system prompt support for the Responses API (#6893) by @Advayp in #6893
  • Constraint violation check is improved to control behavior when violations occur within a batch (#6897) by @phillipleblanc in #6897
  • fix: Multi-column text search with v1/search (#6905) by @peasee in #6905
  • fix: Correctly project text search primary keys to underlying projection (#6904) by @peasee in #6904
  • fix: Update benchmark snapshots (#6901) by @app/github-actions in #6901
  • In S3vector, do not pushdown on non-filterable columns (#6884) by @Jeadie in #6884
  • Run E2E Test CI macOS build on bigger runners (#6896) by @phillipleblanc in #6896
  • Enable configuration of the Responses API for the Azure model provider (#6891) by @Advayp in #6891
  • fix: Update benchmark snapshots (#6888) by @app/github-actions in #6888
  • Update OpenAPI specification for /v1/responses (#6889) by @Advayp in #6889
  • Add test to ensure tools are injected correctly in the Responses API (#6886) by @Advayp in #6886
  • Enable embeddings for append streams (#6878) by @sgrebnov in #6878
  • Show correct limit for EXPLAIN plans in S3VectorsQueryExec (#6852) by @Jeadie in #6852
  • Responses API support for Azure Open AI (#6879) by @Advayp in #6879
  • fix: Update search test case structure (#6865) by @peasee in #6865
  • Fix mongodb benchmark (#6883) by @phillipleblanc in #6883
  • Support multiple column primary keys for S3 vectors. (#6775) by @Jeadie in #6775
  • Kafka Data Connector: persist consumer between restarts (#6870) by @sgrebnov in #6870
  • Fix newlines in errors added in recent PRs (#6877) by @phillipleblanc in #6877
  • Add override parameter to force support for the Responses API (#6871) by @Advayp in #6871
  • Don't use metadata columns in VectorScanTableProvider (#6854) by @Jeadie in #6854
  • Add non-streaming tool call support (hosted and Spice tools) via the Responses API (#6869) by @Advayp in #6869
  • Update error guideline to remove newlines + remove newlines from error messages. (#6866) by @phillipleblanc in #6866
  • Remove void acceleration engine + optional table behaviors (#6868) by @phillipleblanc in #6868
  • Kafka Data Connector basic support (#6856) by @sgrebnov in #6856
  • Federated+Accelerated TPCH Benchmarks for MongoDB (#6788) by @krinart in #6788
  • Pass embeddings calculated in compute_index to the acceleration (#6792) by @phillipleblanc in #6792
  • Add non-streaming and streaming support for OpenAI Responses API endpoint (#6830) by @Advayp in #6830
  • Use latest version of OpenAI crate to resolve issues with Service Tier deserialization (#6853) by @Advayp in #6853
  • Update openapi.json (#6799) by @app/github-actions in #6799
  • Improve management message (#6850) by @lukekim in #6850
  • fix: Include FTS search column if it is the PK (#6836) by @peasee in #6836
  • Refactor Health Checks (#6848) by @Advayp in #6848
  • Introduce a Responses trait and LLM registry for model providers that support the OpenAI Responses API (#6798) by @Advayp in #6798
  • fix: Update datafusion-table-providers to include constraints (#6837) by @peasee in #6837
  • Bump postcard from 1.1.2 to 1.1.3 (#6841) by @app/dependabot in #6841
  • Bump governor from 0.10.0 to 0.10.1 (#6835) by @app/dependabot in #6835
  • Bump ctor from 0.2.9 to 0.5.0 (#6827) by @app/dependabot in #6827
  • Bump azure_core from 0.26.0 to 0.27.0 (#6826) by @app/dependabot in #6826
  • Bump rstest from 0.25.0 to 0.26.1 (#6825) by @app/dependabot in #6825
  • Use latest commit in our fork of async-openai (#6829) by @Advayp in #6829
  • Bump rustls from 0.23.27 to 0.23.31 (#6824) by @app/dependabot in #6824
  • Bump async-trait from 0.1.88 to 0.1.89 (#6823) by @app/dependabot in #6823
  • Bump hyper from 1.6.0 to 1.7.0 (#6814) by @app/dependabot in #6814
  • Bump serde_json from 1.0.140 to 1.0.142 (#6812) by @app/dependabot in #6812
  • Add s3 vector test retrieving vectors (#6786) by @Jeadie in #6786
  • fix: Allow v1/search with only FTS (#6811) by @peasee in #6811
  • Bump tantivy from 0.24.1 to 0.24.2 (#6806) by @app/dependabot in #6806
  • Bump tokio-util from 0.7.15 to 0.7.16 (#6810) by @app/dependabot in #6810
  • fix: Improve FTS index primary key handling (#6809) by @peasee in #6809
  • Bump logos from 0.15.0 to 0.15.1 (#6808) by @app/dependabot in #6808
  • Bump hf-hub from 0.4.2 to 0.4.3 (#6807) by @app/dependabot in #6807
  • Bump odbc-api from 13.0.1 to 13.1.0 (#6803) by @app/dependabot in #6803
  • fix: Spice search CLI with FTS supports string or slice unmarshalling (#6805) by @peasee in #6805
  • Bump uuid from 1.17.0 to 1.18.0 (#6797) by @app/dependabot in #6797
  • Bump reqwest from 0.12.22 to 0.12.23 (#6796) by @app/dependabot in #6796
  • Bump anyhow from 1.0.98 to 1.0.99 (#6795) by @app/dependabot in #6795
  • Bump clap from 4.5.41 to 4.5.45 (#6794) by @app/dependabot in #6794
  • Respect default MAX_DECODING_MESSAGE_SIZE (100MB) in Flight API (#6802) by @sgrebnov in #6802
  • Fix compilation errors caused by upgrading async-openai (#6793) by @Advayp in #6793
  • Remove outdated vector search benchmark (replaced with testoperator) (#6791) by @sgrebnov in #6791
  • Handle errors in vector ingestion pipeline (#6782) by @phillipleblanc in #6782
  • fix: Explicitly error when chunking is defined for vector engines (#6787) by @peasee in #6787
  • Make VectorScanTableProvider and VectorQueryTableProvider support multi-column primary keys (#6757) by @Jeadie in #6757
  • Use megascience/megascience Q+A dataset for text search testing. (#6702) by @Jeadie in #6702
  • Flight REPL autocomplete (#6589) by @krinart in #6589
  • use ref: github.event.pull_request.head.sha in integration_models.yml (#6780) by @Jeadie in #6780
  • fix: Move search telemetry calls in UDTF to scan (#6778) by @peasee in #6778
  • Fix Hugging Face models and embeddings loading in Docker (#6777) by @ewgenius in #6777
  • feat: Migrate bedrock rate limiter (#6773) by @peasee in #6773
  • Run the PR checks on the DEV runners (#6769) by @phillipleblanc in #6769
  • feat: add OpenAI models rate controller (#6767) by @peasee in #6767
  • Implement MongoDB data connector (#6594) by @krinart in #6594
  • fix: Use head ref for concurrency group (#6770) by @peasee in #6770
  • fix: Run enforce pulls with spice on pull_request_target (#6768) by @peasee in #6768
  • feat: Add OpenAI Embeddings Rate Controller (#6764) by @peasee in #6764
  • Move AWS SDK credential bridge integration test to the existing AWS SDK integration test run (#6766) by @phillipleblanc in #6766
  • Use Spice specific errors instead of OpenAIError in embedding module (#6748) by @kczimm in #6748
  • Use context in Glue Catalog Provider (#6763) by @Advayp in #6763
  • pin cargo-deny to previous version (#6762) by @kczimm in #6762
  • Bump actions/download-artifact from 4 to 5 (#6720) by @app/dependabot in #6720
  • Upgrade dependabot dependencies (#6754) by @phillipleblanc in #6754
  • Set E2E Test CI models build to 90 minute timeout (#6756) by @phillipleblanc in #6756
  • chore: upgrade to Rust 1.87.0 (#6614) by @kczimm in #6614
  • feat: Add initial runtime-rate-limiter crate (#6753) by @peasee in #6753
  • feat: Add more embedding traces, add MiniLM MTEB spicepod (#6742) by @peasee in #6742
  • Update QA analytics for release (#6740) by @Advayp in #6740
  • Always use 'returnData: true' for s3 vector query index (#6741) by @Jeadie in #6741
  • feat: Add Embedding and Search anonymous telemetry (#6737) by @peasee in #6737
  • Add 1.5.2 to SECURITY.md (#6739) by @ewgenius in #6739
  • Combine the Iceberg and Object Store AWS SDK bridges into one crate (#6718) by @Advayp in #6718
  • Updates to v1.5.2 release notes (#6736) by @lukekim in #6736
  • Update end game template - move glue catalog to catalogs section (#6732) by @ewgenius in #6732
  • Update v1.5.2.md (#6735) by @kczimm in #6735
  • Add note about S3 Vectors workaround (#6734) by @phillipleblanc in #6734
  • feat: Avoid joining for VectorScanTableProvider if the index is sufficient (#6714) by @peasee in #6714
  • update changelog (#6729) by @kczimm in #6729
  • remove unneeded autogenerated s3 vector code (#6715) by @Jeadie in #6715
  • fix: Set S3 vectors default limit to 30, add more tracing (#6712) by @peasee in #6712
  • docs: Add Hadoop cookbook to endgame template (#6708) by @peasee in #6708
  • Fix testoperator append mode compilation error (#6706) by @phillipleblanc in #6706
  • test: Add VectorScanTableProvider snapshot tests (#6701) by @peasee in #6701
  • feat: Add Hadoop catalog-mode benchmark (#6684) by @peasee in #6684
  • Move shared AWS crates used in bridges to workspace (#6705) by @Advayp in #6705
  • Use installation id to group connections (#6703) by @Advayp in #6703
  • Add Guardrails for AWS bedrock models (#6692) by @Jeadie in #6692
  • Update bedrock keys for CI. (#6693) by @Jeadie in #6693
  • Update acknowledgements (#6690) by @app/github-actions in #6690
  • ROADMAP updates Aug 1, 2025 (#6667) by @lukekim in #6667
  • Add retry logic for OpenAI embeddings creation (#6656) by @sgrebnov in #6656
  • Make models E2E chat test more robust (#6657) by @sgrebnov in #6657
  • Update Search GH Workflow to use Test Operator (#6650) by @sgrebnov in #6650
  • Score and P95 latency calculation for MTEB Quora-based vector search tests in Test Operator (#6640) by @sgrebnov in #6640
  • Fix multiple query error being classified as an internal error (#6635) by @Advayp in #6635
  • Add Support for S3 Table Buckets (#6573) by krinart in #6573
  • set MISTRALRS_METAL_PRECOMPILE=0 for metal (#6652) by @kczimm in #6652
  • Vector search to push down udtf limit argument into logical sort plan (#6636) by @mach-kernel in #6636
  • docs: Update qa_analytics.csv (#6643) by @peasee in #6643
  • Update SECURITY.md (#6642) by @Jeadie in #6642
  • docs: Update qa_analytics.csv (#6641) by @peasee in #6641
  • Separate token usage (#6619) by @Advayp in #6619
  • Fix typo in release notes (#6634) by @Advayp in #6634
  • Add environment variable for org token (#6633) by @Advayp in #6633
  • CDC: Compute embeddings on ingest (#6612) by @mach-kernel in #6612
  • Add view name to view creation errors (#6611) by @lukekim in #6611
  • Add core logic for running MTEB Quora-based vector search tests in Test Operator (#6607) by @sgrebnov in #6607
  • Revert "Update generate-openapi.yml (#6584)" (#6620) by @Jeadie in #6620
  • Non-accelerated views should report as ready only after all dependent datasets are ready (#6617) by @sgrebnov in #6617

Spice v1.5.2 (Aug 11, 2025)

· 7 min read
Kevin Zimmerman
Principal Software Engineer at Spice AI

Announcing the release of Spice v1.5.2! 🛠️

Spice v1.5.2 introduces a new Amazon Bedrock Models Provider for converse API (Nova) compatible models, AWS Redshift support using the Postgres data connector, and Hadoop Catalog Support for Iceberg tables along with several bug fixes and improvements.

What's New in v1.5.2

Amazon Bedrock Models Provider: Adds a new Amazon Bedrock LLM Provider. Models compatible with the Converse API (Nova) are supported.

Amazon Bedrock provides access to a range of foundation models for generative AI. Spice supports using Bedrock-hosted models by specifying the bedrock prefix in the from field and configuring the required parameters.

Supported Model IDs:

  • amazon.nova-lite-v1:0
  • amazon.nova-micro-v1:0
  • amazon.nova-premier-v1:0
  • amazon.nova-pro-v1:0

Refer to the Amazon Bedrock documentation for details on available models and cross-region inference profiles.

Example Spicepod.yaml:

models:
- from: bedrock:us.amazon.nova-lite-v1:0
name: novash
params:
aws_region: us-east-1
aws_access_key_id: ${ secrets:AWS_ACCESS_KEY_ID }
aws_secret_access_key: ${ secrets:AWS_SECRET_ACCESS_KEY }
bedrock_guardrail_identifier: arn:aws:bedrock:abcdefg012927:0123456789876:guardrail/hello
bedrock_guardrail_version: DRAFT
bedrock_trace: enabled
bedrock_temperature: 42

For more information, see the Amazon Bedrock Documentation.

AWS Redshift Support for Postgres Data Connector: Spice now supports connecting to Amazon Redshift using the PostgreSQL data connector. Redshift is a columnar OLAP database compatible with PostgreSQL, allowing you to use the same connector and configuration parameters.

To connect to Redshift, use the format postgres:schema.table in your Spicepod and set the connection parameters to match your Redshift cluster settings.

Example Spicepod.yaml:

# Example datasets for Redshift TPCH tables
datasets:
- from: postgres:public.customer
name: customer
params:
pg_host: ${secrets:PG_HOST}
pg_port: 5439
pg_sslmode: prefer
pg_db: dev
pg_user: ${secrets:PG_USER}
pg_pass: ${secrets:PG_PASS}
- from: postgres:public.lineitem
name: lineitem
params:
pg_host: ${secrets:PG_HOST}
pg_port: 5439
pg_sslmode: prefer
pg_db: dev
pg_user: ${secrets:PG_USER}
pg_pass: ${secrets:PG_PASS}

Redshift types are mapped to PostgreSQL types. See the PostgreSQL connector documentation for details on supported types and configuration.

Hadoop Catalog Support for Iceberg: The Iceberg Data and Catalog connectors now support connecting to Hadoop catalogs on filesystem (file://) or S3 object storage (s3://, s3a://). This enables connecting to Iceberg catalogs without a separate catalog provider service.

Example Spicepod.yaml:

catalogs:
- from: iceberg:file:///tmp/hadoop_warehouse/
name: local_hadoop
- from: iceberg:s3://my-bucket/hadoop_warehouse/
name: s3_hadoop

# Example datasets
- from: iceberg:file:///data/hadoop_warehouse/test/my_table_1
name: local_hadoop
- from: iceberg:s3://my-bucket/hadoop_warehouse/test/my_table_2
name: s3_hadoop

For more details, see the Iceberg Data Connector documentation and the Iceberg Catalog Connector documentation.

Parquet Reader: Optional Parquet Page Index: Fixed an issue where the Parquet reader, using arrow-rs and DataFusion, errored on files missing page indexes, despite the Parquet spec allowing optional indexes. The Spice team contributed optional page index support to arrow-rs (PR #6) and configurable handling in DataFusion (PR #93). A new runtime parameter, parquet_page_index, makes Parquet Page Indexes configurable in Spice:

runtime:
params:
parquet_page_index: required # Options: required, skip, auto
  • required: (Default) Errors if page indexes are absent.
  • skip: Ignores page indexes, potentially reducing query performance.
  • auto: Uses page indexes if available; skips otherwise.

This improves compatibility and query flexibility for Parquet datasets.

Contributors

Breaking Changes

Amazon S3 Vectors Vector Engine: Amazon S3 Vectors is currently a preview AWS service. A recent update to the Amazon S3 Vectors service API introduced a breaking change that affects the integration when projecting (selecting) the embedding column. This results in the following error:

Json error: whilst decoding field 'data': expected [ got nullReceived only partial JSON payload from QueryVectors

The issue is expected to be resolved in the next release of Spice. A current workaround is to limit queries to non-embedding columns.

i.e. instead of:

SELECT url, title, scored, body_embedding
FROM vector_search(pulls, 'bugs in DuckDB', 4)
WHERE state = 'OPEN'
ORDER BY score DESC
LIMIT 4;

Remove the *_embedding column from the projection. E.g.

SELECT url, title, scored
FROM vector_search(pulls, 'bugs in DuckDB', 4)
WHERE state = 'OPEN'
ORDER BY score DESC
LIMIT 4;

This issue and workaround also applies to SELECT * FROM vector_search(..). E.g.

SELECT *
FROM vector_search(pulls, 'bugs in DuckDB', 4)
WHERE state = 'OPEN'
ORDER BY score DESC
LIMIT 4;

Cookbook Updates

The Spice Cookbook includes 75 recipes to help you get started with Spice quickly and easily.

Upgrading

To upgrade to v1.5.2, use one of the following methods:

CLI:

spice upgrade

Homebrew:

brew upgrade spiceai/spiceai/spice

Docker:

Pull the spiceai/spiceai:1.5.2 image:

docker pull spiceai/spiceai:1.5.2

For available tags, see DockerHub.

Helm:

helm repo update
helm upgrade spiceai spiceai/spiceai

AWS Marketplace:

🎉 Spice is also now available in the AWS Marketplace!

What's Changed

Dependencies

No major dependency updates.

Changelog

  • fixes for databricks OpenAI compatibility (#6629) by @Jeadie in #6629
  • Update spicepod.schema.json (#6632) by @app/github-actions in #6632
  • Remove 'stream_options' from databricks LLMs (#6637) by @Jeadie in #6637
  • Move retry and rate limiting logic for Amazon bedrock out of embeddings. (#6626) by @Jeadie in #6626
  • Disable Metal precomplation in integration_llms.yml (#6649) by @Jeadie in #6649
  • fix: Hadoop integration test (#6660) by @peasee in #6660
  • feat: Add Hadoop Catalog Data Component (#6658) by @peasee in #6658
  • update datafusion-table-providers to latest spiceai tag (#6661) by @mach-kernel in #6661
  • feat: Add Hadoop Catalog connectors for Iceberg (#6659) by @peasee in #6659
  • Make FullTextSearchExec robust to RecordBatch column ordering. (#6675) by @Jeadie in #6675
  • Make 'runtime-object-store' crate (#6674) by @Jeadie in #6674
  • fix: Support include for Iceberg (#6663) by @peasee in #6663
  • feat: Add Hadoop TPCH benchmark (#6678) by @peasee in #6678
  • feat: Add Hadoop metadata_path parameter (#6680) by @peasee in #6680
  • fix: Automatically infer Hadoop warehouse scheme (#6681) by @peasee in #6681
  • Amazon Bedrock, specifically Nova models (#6673) by @Jeadie in [#6673](https://github.com/spiceai/spiceai/pull/6673
  • fix perplexity_auth_token parameters for web_search (#6685) by @Jeadie in #6685
  • Fix AWS Auth issue (#6699) by @Advayp in #6699
  • Limit Concurrent Requests for GitHub (#6672) by @Advayp in #6672
  • Add runtime parameter to enable more permissive parquet reading when page indexes are missing (#6716) by @phillipleblanc in #6716
  • Improve Flight REPL error messages (#6696) by @lukekim in #6696
  • Fixes from search tests (#6710) by @Jeadie in #6710

Amazon S3 Vectors with Spice

· 26 min read
Jack Eadie
Token Plumber at Spice AI

The latest Spice.ai Open Source release (v1.5.0) brings major improvements to search, including native support for Amazon S3 Vectors. Announced in public preview at AWS Summit New York 2025, Amazon S3 Vectors is a new S3 bucket type purpose-built for vector embeddings, with dedicated APIs for similarity search.

Spice AI was a day 1 launch partner for S3 Vectors, integrating it as a scalable vector index backend. In this post, we explore how S3 Vectors integrates into Spice.ai’s data, search, and AI-inference engine, how Spice manages indexing and lifecycle of embeddings for production vector search, and how this unlocks a powerful hybrid search experience. We’ll also put this in context with industry trends and compare Spice’s approach to other vector database solutions like Qdrant, Weaviate, Pinecone, and Turbopuffer.

Amazon S3 Vectors Overview

Amazon S3 Vectors Overview

Amazon S3 Vectors extends S3 object storage with native support for storing and querying vectors at scale. As AWS describes, it is “designed to provide the same elasticity, scale, and durability as Amazon S3,” providing storage of billions of vectors and sub-second similarity queries. Crucially, S3 Vectors dramatically lowers the cost of vector search infrastructure – reducing upload, storage, and query costs by up to 90% compared to traditional solutions. It achieves this by separating storage from compute: vectors reside durably in S3, and queries execute on transient, on-demand resources, avoiding the need for always-on, memory-intensive vector database servers. In practice, S3 Vectors exposes two core operations:

  1. Upsert vectors – assign a vector (an array of floats) to a given key (identifier) and optionally store metadata alongside it.

  2. Vector similarity query – given a new query vector, efficiently find the stored vectors that are closest (e.g. minimal distance) to it, returning their keys (and scores).

This transforms S3 into a massively scalable vector index service. You can store embeddings at petabyte scale and perform similarity search with metrics like cosine or Euclidean distance via a simple API. It’s ideal for AI use cases like semantic search, recommendations, or Retrieval-Augmented Generation (RAG) where large volumes of embeddings need to be queried semantically. By leveraging S3’s pay-for-use storage and ephemeral compute, S3 Vectors can handle infrequent or large-scale queries much more cost-effectively than memory-bound databases, yet still deliver sub-second results.

Vector Search with Embeddings

Vector similarity search retrieves data by comparing items in a high-dimensional embedding space rather than by exact keywords. In a typical pipeline:

  • Data to vectors: We first convert each data item (text, image, etc.) into a numeric vector representation (embedding) using an ML model. For example, a customer review text might be turned into a 768-dimensional embedding that encodes its semantic content. Models like Amazon Titan Embeddings, OpenAI, or Hugging Face sentence transformers handle this step.

  • Index storage: These vectors are stored in a specialized index or database optimized for similarity search. This could be a dedicated vector database or, in our case, Amazon S3 Vectors acting as the index. Each vector is stored with an identifier (e.g. the primary key of the source record) and possibly metadata.

  • Query by vector: A search query (e.g. a phrase or image) is also converted into an embedding vector. The vector index is then queried to find the closest stored vectors by distance metric (cosine, Euclidean, dot product, etc.). The result is a set of IDs of the most similar items, often with a similarity score.

This process enables semantic search – results are returned based on meaning and similarity rather than exact text matches. It powers features like finding relevant documents by topic even if exact terms differ, recommendation systems (finding similar user behavior or content), and providing knowledge context to LLMs in RAG. With the Spice.ai Open Source integration, this whole lifecycle (embedding data, indexing vectors, querying) is managed by the Spice runtime and exposed via a familiar SQL or HTTP interface.

Amazon S3 Vectors in Spice.ai

Spice integration with Amazon S3 Vectors

Spice.ai is an open-source data, search and AI compute engine that supports vector search end-to-end. By integrating S3 Vectors as an index, Spice can embed data, store embeddings in S3, and perform similarity queries – all orchestrated through simple configuration and SQL queries. Let’s walk through how you enable and use this in Spice.

Configuring a Dataset with Embeddings

To use vector search, annotate your dataset schema to specify which column(s) to embed and with which model. Spice supports various embedding models (both local or hosted) via the embeddings section in the configuration. For example, suppose we have a customer reviews table and we want to enable semantic search over the review text (body column):

datasets:
- from: oracle:"CUSTOMER_REVIEWS"
name: reviews
columns:
- name: body
embeddings:
from: bedrock_titan # use an embedding model defined below

embeddings:
- from: bedrock:amazon.titan-embed-text-v2:0
name: bedrock_titan
params:
aws_region: us-east-2
dimensions: '256'

In this spicepod.yaml, we defined an embedding model bedrock_titan (in this case AWS's Titan text embedding model) and attached it to the body column. When the Spice runtime ingests the dataset, it will automatically generate a vector embedding for each row’s body text using that model. By default, Spice can either store these vectors in its acceleration layer or compute them on the fly. However, with S3 Vectors, we can offload them to an S3 Vectors index for scalable storage.

To use S3 Vectors, we simply enable the vector engine in the dataset config:

datasets:
- from: oracle:"CUSTOMER_REVIEWS"
name: reviews
vectors:
enabled: true
engine: s3_vectors
params:
s3_vectors_bucket: my-s3-vector-bucket
#... (rest of dataset definition as above)

This tells Spice to create or use an S3 Vectors index (in the specified S3 bucket) for storing the body embeddings. Spice manages the entire index lifecycle: it creates the vector index, handles inserting each vector with its primary key into S3, and knows how to query it. The embedding model and data source are as before – the only change is where the vectors are stored and queried. The benefit is that now our vectors reside in S3’s highly scalable storage, and we can leverage S3 Vectors’ efficient similarity search API.

Performing a Vector Search Query

Once configured, performing a semantic search is straightforward. Spice exposes both an HTTP endpoint and a SQL table-valued function for vector search. For example, using the HTTP API:

curl -X POST http://localhost:8090/v1/search \
-H "Content-Type: application/json" \
-d '{
"datasets": ["reviews"],
"text": "issues with same day shipping",
"additional_columns": ["rating", "customer_id"],
"where": "created_at >= now() - INTERVAL '7 days'",
"limit": 2
}'

This JSON query says: search the reviews dataset for items similar to the text "issues with same day shipping", and return the top 2 results, including their rating and customer id, filtered to reviews from the last 7 days. The Spice engine will embed the query text (using the same model as the index), perform a similarity lookup in the S3 Vectors index, filter by the WHERE clause, and return the results. A sample response might look like:

{
"results": [
{
"matches": {
"body": "Everything on the site made it seem like I’d get it the same day. Still waiting the next morning was a letdown."
},
"data": { "rating": 3, "customer_id": 6482 },
"primary_key": { "review_id": 123 },
"score": 0.82,
"dataset": "reviews"
},
{
"matches": {
"body": "It was marked as arriving 'today' when I paid, but the delivery was pushed back without any explanation. Timing was kind of important for me."
},
"data": { "rating": 2, "customer_id": 3310 },
"primary_key": { "review_id": 24 },
"score": 0.76,
"dataset": "reviews"
}
],
"duration_ms": 86
}

Each result includes the matching column snippet (body), the additional requested fields, the primary key, and a relevance score. In this case, the two reviews shown are indeed complaints about “same day” delivery issues, which the vector search found based on semantic similarity to the query (see how the second result made no mention of "same day" delivery, but rather described a similar issue as the first ).

Developers can also use SQL for the same operation. Spice provides a table function vector_search(dataset, query) that can be used in the FROM clause of a SQL query. For example, the above search could be expressed as:

SELECT review_id, rating, customer_id, body, score
FROM vector_search(reviews, 'issues with same day shipping')
WHERE created_at >= to_unixtime(now() - INTERVAL '7 days')
ORDER BY score DESC
LIMIT 2;

This would yield a result set (with columns like review_id, score, etc.) similar to the JSON above, which you can join or filter just like any other SQL table. This ability to treat vector search results as a subquery/table and combine them with standard SQL filtering is a powerful feature of Spice.ai’s integration – few other solutions let you natively mix vector similarity and relational queries so seamlessly.

See a 2-min demo of it in action:

Managing Embeddings Storage in Spice.ai

An important design question for any vector search system is where and how to store the embedding vectors. Before introducing S3 Vectors, Spice offered two approaches for managing vectors:

  1. Accelerator storage: Embed the data in advance and store the vectors alongside other cached data in a Data Accelerator (Spice’s high-performance materialization layer). This keeps vectors readily accessible in memory or fast storage.

  2. Just-in-time computation: Compute the necessary vectors on the fly during a query, rather than storing them persistently. For example, at query time, embed only the subset of rows that satisfy recent filters (e.g. all reviews in the last 7 days) and compare those to the query vector.

Both approaches have trade-offs. Pre-storing in an accelerator provides fast query responses but may not be feasible for very large datasets (which might not fit entirely, or fit affordably in fast storage) and accelerators, like DuckDB or SQLite aren’t optimized for similarity search algorithms on billion-scale vectors. Just-in-time embedding avoids extra storage but becomes prohibitively slow when computing embeddings over large data scans (and for each query), and provides no efficient algorithm for efficiently finding similar neighbours.

Amazon S3 Vectors offers a compelling third option: the scalability of S3 with the efficient retrieval of vector index data structures. By configuring the dataset with engine: s3_vectors as shown earlier, Spice will offload the vector storage and similarity computations to S3 Vectors. This means you can handle very large embedding sets (millions or billions of items) without worrying about Spice’s memory or local disk limits, and still get fast similarity operations via S3’s API. In practice, when Spice ingests data, it will embed each row’s body and PUT it into the S3 Vector index (with the review_id as the key, and possibly some metadata). At query time, Spice calls S3 Vectors’ query API to retrieve the nearest neighbors for the embedded query. All of this is abstracted away; you simply query Spice and it orchestrates these steps.

The Spice runtime manages index creation, updates, and deletion. For instance, if new data comes in or old data is removed, Spice will synchronize those changes to the S3 vector index. Developers don’t need to directly interact with S3 – it’s configured once in YAML. This tight integration accelerates application development: your app can treat Spice like any other database, while behind the scenes Spice leverages S3’s elasticity for the heavy lifting.

Vector Index Usage in Query Execution

How does a vector index actually get used in Spice’s SQL query planner? To illustrate, consider the simplified SQL we used:

SELECT *
FROM vector_search(reviews, 'issues with same day shipping')
ORDER BY score DESC
LIMIT 5;

Logically, without a vector index, Spice would have to do the following at query time:

  1. Embed the query text 'issues with same day shipping' into a vector v.

  2. Retrieve or compute all candidate vectors for the searchable column (here every body embedding in the dataset). This could mean scanning every row or at least every row matching other filter predicate.

  3. Calculate distances between the query vector v and each candidate vector, compute a similarity score (e.g. score = 1 - distance).

  4. Sort all candidates by the score and take the top 5.

For large datasets, steps 2–4 would be extremely expensive (a brute-force scan through potentially millions of vectors for each search, then a full sort operation). A vector index avoiding unnecessary recomputation of embeddings, reduces the number of distance calculations required, and provides in-order candidate neighbors.

With S3 Vectors, step 2 and 3 are pushed down to the S3 service. The vector index can directly return the top K closest matches to v. Conceptually, S3 Vectors gives back an ordered list of primary keys with their similarity scores. For example, it might return something like: {(review_id=123, score=0.82), (review_id=24, score=0.76), ...} up to K results.

Spice then uses these results, logically as a temporary table (let’s call it vector_query_results), joined with the main reviews table to get the full records. In SQL pseudocode, Spice does something akin to:

-- The vector index returns the closest matches for a given query.
CREATE TEMP TABLE vector_query_results (
review_id BIGINT,
score FLOAT
);

Imagine this temp table is populated by an efficient vector retrieval operatin in S3 Vectors for the query.

-- Now we join to retrieve full details
SELECT r.review_id, r.rating, r.customer_id, r.body, v.score
FROM vector_query_results v
JOIN reviews r ON r.review_id = v.review_id
ORDER BY v.score DESC
LIMIT 5;

This way, only the top few results (say 50 or 100 candidates) are processed in the database, rather than the entire dataset. The heavy work of narrowing down candidates occurs inside the vector index. Spice essentially treats vector_search(dataset, query) as a table-valued function that produces (id, score) pairs which are then joinable.

Handling Filters Efficiently

One consideration when using an external vector index is how to handle additional filter conditions (the WHERE clause). In our example, we had a filter created_at >= now() - 7 days. If we simply retrieve the top K results from the vector search and then apply the time filter, we might run into an issue: those top K might not include any recent items, even if there are relevant recent items slightly further down the similarity ranking. This is because S3 Vectors (like most ANN indexes) will return the top K most similar vectors globally, unaware of our date constraint.

If only a small fraction of the data meets the filter, a naive approach could drop most of the top results, leaving fewer than the desired number of final results. For example, imagine the vector index returns 100 nearest reviews overall, but only 5% of all reviews are from the last week – we’d expect only ~5 of those 100 to be recent, possibly fewer than the LIMIT. The query could end up with too few results not because they don’t exist, but because the index wasn’t filter-aware and we truncated the candidate list.

To solve this, S3 Vectors supports metadata filtering at query time. We can store certain fields as metadata with each vector and have the similarity search constrained to vectors where the metadata meets criteria. Spice.ai leverages this by allowing you to mark some dataset columns as “vector filterable”. In our YAML, we could do:

columns:
- name: created_at
metadata:
vectors: filterable

By doing this, Spice's query planner will include the created_at value with each vector it upserts to S3, and it will push down the time filter into the S3 Vectors query. Under the hood, the S3 vector query will then return only nearest neighbors that also satisfy created_at >= now()-7d. This greatly improves both efficiency and result relevance. The query execution would conceptually become:

-- Vector query with filter returns a temp table including the metadata
CREATE TEMP TABLE vector_query_results (
review_id BIGINT,
score FLOAT,
created_at TIMESTAMP
);
-- vector_query_results is already filtered to last 7 days

SELECT r.review_id, r.rating, r.customer_id, r.body, v.score
FROM vector_query_results v
JOIN reviews r ON r.review_id = v.review_id
-- (no need for additional created_at filter here, it’s pre-filtered)
ORDER BY v.score DESC
LIMIT 5;

Now the index itself is ensuring all similar reviews are from the last week, and so if there are at least five results from the last week, it will return a full result (i.e. respecting LIMIT 5).

Including Data to Avoid Joins

Another optimization Spice supports is storing additional, non-filterable columns in the vector index to entirely avoid the expensive table join back to the main table for certain queries. For example, we might mark rating, customer_id, or even the text body as non-filterable vector metadata. This means these fields are stored with the vector in S3, but not used for filtering (just for retrieval). In the Spice config, it would look like:

columns:
- name: rating
metadata:
vectors: non-filterable
- name: customer_id
metadata:
vectors: non-filterable
- name: body
metadata:
vectors: non-filterable

With this setup, when Spice queries S3 Vectors, the vector index will return not only each match’s review_id and score, but also the stored rating, customer_id, and body values. Thus, the temporary vector_query_results table already has all the information needed to satisfy the query. We don’t even need to join against the reviews table unless we want some column that wasn’t stored. The query can be answered entirely from the index data:

SELECT review_id, rating, customer_id, body, score
FROM vector_query_results
ORDER BY score DESC
LIMIT 5;

This is particularly useful for read-heavy query workloads where hitting the main database adds latency. By storing the most commonly needed fields along with the vector, Spice’s vector search behaves like an index-only query (similar to covering indexes in relational databases). You trade a bit of extra storage in S3 (duplicating some fields, but still managed by Spice) for faster queries that bypass the heavier join.

This extends to WHERE conditions on non-filterable columns, or filter predicate unsupported by S3 vectors. Spice's execution engine can apply these filters, still avoiding any expensive JOIN on the underlying table.

SELECT review_id, rating, customer_id, body, score
FROM vector_query_results
where rating > 3 -- Filter performed in Spice on, with non-filterable data from vector index
ORDER BY score DESC
LIMIT 5;

It’s worth noting that you should choose carefully which fields to mark as metadata – too many or very large fields could increase index storage and query payload sizes. Spice gives you the flexibility to include just what you need for filtering and projection to optimize each use case.

Beyond Basic Vector Search in Spice

Many real-world search applications go beyond a single-vector similarity lookup. Spice.ai’s strength is that it’s a full database engine. You can compose more complex search workflows, including hybrid search (combining keyword/text search with vector search), multi-vector queries, re-ranking strategies, and more. Spice provides both an out-of-the-box hybrid search API and the ability to write custom SQL to implement advanced retrieval logic.

  • Multiple vector fields or multi-modal search: You might have vectors for different aspects of data (e.g. an e-commerce product could have embeddings for both its description and the product's image. Or a document has both a title and body that should be searchable individually and together) that you may want to search across and combine results. Spice lets you do vector search on multiple columns easily, and you can weight the importance of each. For instance, you might boost matches in the title higher than matches in the body.

  • Vector and full-text search: Similar to vector search, columns can have text indexes defined that enable full-text BM25 search. Text search can then be performed in SQL with a similar text_search UDTF. The /v1/search HTTP API will perform a hybrid search across both full-text and vector indexes, merging results using Reciprocal Rank Fusion (RRF). This means you get a balanced result set that accounts for direct keyword matches as well as semantic similarity. The example below demonstrates how RRF can be implemented in SQL by combining ranks.

  • Hybrid vector + keyword search: Sometimes you want to ensure certain keywords are present while also using semantic similarity. Spice supports hybrid search natively – its default /v1/search HTTP API actually performs both full-text BM25 search and vector search, then merges results using Reciprocal Rank Fusion (RRF). This means you get a balanced result set that accounts for direct keyword matches as well as semantic similarity. In Spice’s SQL, you can also call text_search(dataset, query) for traditional full-text search, and combine it with vector_search results. The example below demonstrates how RRF can be implemented in SQL by combining ranks.

  • Two-phase retrieval (re-ranking): A common pattern is to use a fast first-pass retrieval (e.g. a keyword search) to get a larger candidate set, then apply a more expensive or precise ranking (e.g. vector search) on this subset to improve the score of the required final candidate set. With Spice, you can orchestrate this in SQL or in application code. For example, you could query a BM25 index for 100 candidates, then perform a vector search amongst this candidate set(i.e. restricted to those IDs) for a second phase. Since Spice supports standard SQL constructs, you can express these multi-step plans with common table expressions (CTEs) and joins.

To illustrate hybrid search, here’s a SQL snippet that uses the Reciprocal Rank Fusion (RRF) technique to merge vector and text search results for the same query (RRF is used, when needed, in the v1/search HTTP API):

WITH
vector_results AS (
SELECT review_id, RANK() OVER (ORDER BY score DESC) AS vector_rank
FROM vector_search(reviews, 'issues with same day shipping')
),
text_results AS (
SELECT review_id, RANK() OVER (ORDER BY score DESC) AS text_rank
FROM text_search(reviews, 'issues with same day shipping')
)
SELECT
COALESCE(v.review_id, t.review_id) AS review_id,
-- RRF scoring: 1/(60+rank) from each source
(1.0 / (60 + COALESCE(v.vector_rank, 1000)) +
1.0 / (60 + COALESCE(t.text_rank, 1000))) AS fused_score
FROM vector_results v
FULL OUTER JOIN text_results t ON v.review_id = t.review_id
ORDER BY fused_score DESC
LIMIT 50;

This takes the vector similarity results and text (BM25) results, assigns each a rank based not on the score, but rather the relative order of candidates, and combines these ranks for an overall order. Spice’s primary key SQL semantics easily enables this document ID join.

For a multi-column vector search example, suppose our reviews dataset has both a title and body with embeddings, and we want to prioritize title matches higher. We could create a combined_score where the title is weighted twice as high as the body:

WITH
body_results AS (
SELECT review_id, score AS body_score
FROM vector_search(reviews, 'issues with same day shipping', col => 'body')
),
title_results AS (
SELECT review_id, score AS title_score
FROM vector_search(reviews, 'issues with same day shipping', col => 'title')
)
SELECT
COALESCE(body.review_id, title.review_id) AS review_id,
COALESCE(body_score, 0) + 2.0 * COALESCE(title_score, 0) AS combined_score
FROM body_results
FULL OUTER JOIN title_results ON body_results.review_id = title_results.review_id
ORDER BY combined_score DESC
LIMIT 5;

These examples scratch the surface of what you can do by leveraging Spice’s SQL-based composition. The key point is that Spice isn’t just a vector database – it’s a hybrid engine that lets you combine vector search with other query logic (text search, filters, joins, aggregations, etc.) all in one place. This can significantly simplify building complex search and AI-driven applications.

(Note: Like most vector search systems, S3 Vectors uses an approximate nearest neighbor (ANN) algorithm under the hood for performance. This yields fast results that are probabilistically the closest, which is usually an acceptable trade-off in practice. Additionally, in our examples we focused on one embedding per row; production systems may use techniques like chunking text into multiple embeddings or adding external context, but the principles above remain the same.)

Industry Context and Comparisons

The rise of vector databases over the past few years (Pinecone, Qdrant, Weaviate, etc.) has been driven by the need to serve AI applications with semantic search at scale. Each solution takes a slightly different approach in architecture and trade-offs. Spice.ai’s integration with Amazon S3 Vectors represents a newer trend in this space: decoupling storage from compute for vector search, analogous to how data warehouses separated compute and storage in the past. Let’s compare this approach with some existing solutions:

  • Traditional Vector Databases (Qdrant, Weaviate, Pinecone): These systems typically run as dedicated services or clusters that handle both the storage of vectors (on disk or in-memory) and the computation of similarity search. For example, Qdrant (an open-source engine in Rust) allows either in-memory storage or on-disk storage (using RocksDB) for vectors and payloads. It’s optimized for high performance and offers features like filtering, quantization, and distributed clustering, but you generally need to provision servers/instances that will host all your data and indexes. Weaviate, another popular open-source vector DB, uses a Log-Structured Merge (LSM) tree based storage engine that persists data to disk and keeps indexes in memory. Weaviate supports hybrid search (it can combine keyword and vector queries) and offers a GraphQL API, with a managed cloud option priced mainly by data volume. Pinecone, a fully managed SaaS, also requires you to select a service tier or pod which has certain memory/CPU allocated for your index – essentially your data lives in Pinecone’s infrastructure, not in your AWS account. These solutions excel at low-latency search for high query throughput scenarios (since data is readily available in RAM or local SSD), but the cost can be high for large datasets. You pay for a lot of infrastructure to be running, even during idle times. In fact, prior to S3 Vectors, vector search engines often stored data in memory at ~$2/GB and needed multiple replicas on SSD, which is “the most expensive way to store data”, as Simon Eskildsen (Turbopuffer’s founder) noted. Some databases mitigate cost by compressing or offloading to disk, but still, maintaining say 100 million embeddings might require a sizable cluster of VMs or a costly cloud plan.

  • Spice.ai with Amazon S3 Vectors: This approach flips the script by storing vectors in cheap, durable object storage (S3) and loading/indexing them on demand. As discussed, S3 Vectors keeps the entire vector dataset in S3 at ~$0.02/GB storage , and only spins up transient compute (managed by AWS) to serve queries, meaning you aren’t paying for idle GPU or RAM time. AWS states this design can cut total costs by up to 90% while still giving sub-second performance on billions of vectors. It’s essentially a serverless vector search model – you don’t manage servers or even dedicated indices; you just use the API. Spice.ai’s integration means developers get this cost-efficiency without having to rebuild their application: they can use standard SQL and Spice will push down operations to S3 Vectors as appropriate. This decoupled storage/compute model is ideal for use cases where the data is huge but query volumes are moderate or bursty (e.g., an enterprise semantic search that is used a few times an hour, or a nightly ML batch job). It avoids the “monolithic database” scenario of having a large cluster running 24/7. However, one should note that if you need extremely high QPS (thousands of queries per second at ultra-low latency), a purely object-storage-based solution might not outperform a tuned in-memory vector DB – AWS positions S3 Vectors as complementary to higher-QPS solutions like OpenSearch for real-time needs.

  • Turbopuffer: Turbopuffer is a startup that, much like Spice with S3 Vectors, is built from first principles on object storage. It provides “serverless vector and full-text search… fast, 10× cheaper, and extremely scalable,” by leveraging S3 or similar object stores with smart caching. The philosophy is the same: use the durability and low cost of object storage for the bulk of data, and layer a cache (memory/SSD) in front for performance-critical portions. According to Turbopuffer’s founder, moving from memory/SSD-centric architectures to an object storage core can yield 100× cost savings for cold data and 6–20× for warm data, without sacrificing too much performance. Turbopuffer’s engine indexes data incrementally on S3 and uses caching to achieve similar latency to conventional search engines on hot data. The key difference is that Turbopuffer is a standalone search service (with its own API), whereas Spice uses AWS’s S3 Vectors service as the backend. Both approaches validate the industry trend toward disaggregated storage for search. Essentially, they are bringing the cloud data warehouse economics to vector search: store everything cheaply, compute on demand.

In summary, Spice.ai’s integration with S3 Vectors and similar efforts indicate a shift in vector search towards cost-efficient, scalable architectures that separate the concerns of storing massive vector sets and serving queries. Developers now have options: if you need blazing fast, realtime vector search with constant high traffic, dedicated compute infrastructure might be justified. But for many applications – enterprise search, AI assistants with a lot of knowledge but lower QPS, periodic analytics over embeddings – offloading to something like S3 Vectors can save enormously on cost while still delivering sub-second performance at huge scale. And with Spice.ai, you get the best of both worlds: the ease of a unified SQL engine that can do keyword + vector hybrid search on structured data, combined with the power of a cloud-native vector store. It simplifies your stack (no separate vector DB service to manage) and accelerates development since you can join and filter vector search results with your data immediately in one query .

References:

Spice v1.5.1 (July 28, 2025)

· 5 min read
Jack Eadie
Token Plumber at Spice AI

Announcing the release of Spice v1.5.1! 🔑

Spice v1.5.1 expands the GitHub data connector to include pull-request comments, adds a configurable rate limiting for AWS Bedrock embedding models, expands partition pruning with inequality operators, and adds client-supplied cache keys for granular caching control in the HTTP and Arrow Flight SQL APIs.

What's New in v1.5.1

GitHub Data Connector Pull Request Comments: Configure GitHub pulls datasets to include comments.

Example Spicepod.yaml:

datasets:
- from: github:github.com/spiceai/spiceai/pulls
name: spiceai.pulls
params:
github_include_comments: all # 'review', 'discussion', or 'none'. Defaults to 'none'.
github_max_comments_fetched: '25' # Defaults to 100
# ...

For details, see the GitHub Data Connector documentation.

AWS Bedrock Embedding Models Invocation Control: Improved rate limiting control for AWS Bedrock embedding models with max_concurrent_invocations configuration.

embeddings:
- from: bedrock:cohere.embed-english-v3
name: cohere-embeddings
params:
max_concurrent_invocations: '41'
# ...

For details, see the AWS Bedrock Embeddings Model Provider documentation.

Improved Query Partitioning: Expanded partition pruning support with additional inequality operators (e.g. >, >=, <, <=).

For details, see the Query Partitioning documentation.

Client-Supplied Cache Keys: Support for a new Spice-Cache-Key header/metadata-key in the HTTP and Arrow Flight SQL query APIs to for fine-grained client-side caching control.

Example HTTP API usage:

$ curl -vvS -XPOST http://localhost:8090/v1/sql \
-H"spice-cache-key: 1851400_20170216_north_america" \
-d "select * from scihub_journals_accessed
where user_id = '1851400'
and date_trunc('DAY', timestamp) = '2017-02-16'
and city = 'New York';"

Example Response:

< HTTP/1.1 200 OK
< content-type: application/json
< x-cache: Hit from spiceai
< results-cache-status: HIT
< vary: Spice-Cache-Key
< vary: origin, access-control-request-method, access-control-request-headers
< content-length: 604
< date: Wed, 23 Jul 2025 20:26:12 GMT
<
[{
"timestamp": "2017-02-16 09:55:06",
"doi": "10.1155/2012/650929",
"ip_identifier": 1000856,
"user_id": 1851400,
"country": "United States",
"city": "New York",
"longitude": 40.7830603,
"latitude": -73.9712488
},
...
]

For details, see the Cache Control documentation.

Contributors

New Contributors

Breaking Changes

  • N/A

Cookbook Updates

No new recipes added in this release.

The Spice Cookbook includes 74 recipes to help you get started with Spice quickly and easily.

Upgrading

To upgrade to v1.5.1, use one of the following methods:

CLI:

spice upgrade

Homebrew:

brew upgrade spiceai/spiceai/spice

Docker:

Pull the spiceai/spiceai:1.5.1 image:

docker pull spiceai/spiceai:1.5.1

For available tags, see DockerHub.

Helm:

helm repo update
helm upgrade spiceai spiceai/spiceai

What's Changed

Dependencies

No major dependency updates.

Changelog

  • Fix refresh via Api when dataset is already accelerated and no refresh interval is set by @sgrebnov in #6549
  • Add support for custom GraphQL unnesting behavior by @Advayp in #6540
  • Regex Update to disallow hyphens dataset names by @varunguleriaCodes in #6383
  • Enforce max limit on comments fetched per PR by @Advayp in #6580
  • Fix accelerated refresh issue by @Advayp in #6590
  • Enable configurations of max invocations for Bedrock models by @Advayp in #6592
  • Client-supplied cache keys (Spice-Cache-Key) by @mach-kernel in #6579
  • Improved partition pruning by @kczimm in #6582
  • Fix retention filter when both retention_sql and period are set by @sgrebnov in #6595
  • Initial support for PR comments by @Advayp in #6569
  • chore: Update croner by @peasee in #6547
  • fix databricks streaming for Claude model by @peasee in #6601
  • Remove FullTextUDTFAnalyzerRule and move FTS code into search crate by @jeadie in #6596
  • Remove download of legacy sentence transformers config by @jeadie in #6605
  • re-add snapshot tests by @jeadie
  • Embedding column config to support client-specified vector sizes by @mach-kernel in #6610
  • Fix mismatch in columns for the GitHub PR table type by @Advayp in #6616
  • bump version to 1.5.1 by @phillipleblanc
  • fix issues with cherry-picking by @jeadie
  • Add integration tests for GitHub PRs with comments by @Advayp in #6581
  • Add view name to view creation errors by @lukekim in #6611
  • CDC: Compute embeddings on ingest by @mach-kernel in #6612