PyMilvus v2.6.6 Release Notes

Features

1. Add comprehensive benchmarking framework (#3151)

2. Add Function SDK support (#3080)

Enhancements

1. Change highlight query keyword (#3143)

Bug Fixes

1. fix: Keep boolean serialization as uppercase for e2e compatibility (#3168)
2. fix: Preserve False boolean values in SCANN index parameters (#3165)
3. fix: Use json to parse deep recursive json when orjson failed (#3156)
4. fix: Fix search params in orm search iterator (#3162)
5. fix: Fix the negative index access issue when preloading float vectors (#3144)
6. fix: Miss to save ts for iterator checkpoint (#3149)
7. fix: Fix some typos and ignore emacs backup files (#3145)

What's Changed

• [Backport 2.6] fix: Keep boolean serialization as uppercase for e2e compatibility (#3168) by @jac0626 in #3168
• [Backport 2.6] fix: preserve False boolean values in SCANN index parameters (#3165) by @jac0626 in #3165
• fix: [2.6] Fix search params in orm search iterator (#3162) by @cai-zhang in #3162
• [Backport 2.6] fix:use json to prase deep recursive json when orjson failed (#3156) by @jac0626 in #3156
• [Backport 2.6] feat(perf): Add comprehensive benchmarking framework (#3151) by @XuanYang-cn in #3151
• fix: miss to save ts for iterator checkpoint (#3149) by @MrPresent-Han in #3149
• [Backport 2.6] Propose fix some typos and ignore emacs backup files (#3145) by @jeis4wpi in #3145
• [Backport 2.6] fix: Fix the negative index access issue when preloading float vectors. (#3144) by @maocatooo in #3144
• [Backport 2.6] enhance: change highlight query keyword (#3143) by @aoiasd in #3143
• [cp 2.6] function sdk (#3080) by @junjiejiangjjj in #3080

Full Changelog: v2.6.5...2.6

PyMilvus v2.6.5 Release Notes

PyMilvus 2.6.5 Release Notes

Enhancements

1. Added comprehensive features to AsyncMilvusClient including: (#3108)

2. Enhanced error messages when AsyncMilvusClient fails to connect. (#3102)

Bug Fixes

1. fix: AsyncMilvusClient.search supports EmbeddingList (#3101)
2. fix: AsyncMilvusClient.create_index supports nested fields (#3100)
3. fix: Use deepcopy for struct_schema in add_field (#3105)
4. fix: Error when function or function_score fetch empty params (#3119)

What's Changed

• Breaking change: Rename Stage to Volume by @lentitude2tk in #3091
• Support timestamptz for bulkwriter by @yhmo in #3114
• [Backport 2.6] fix: AsyncMilvusClient.create_index support nested fields (#3100) by @pymilvus-bot in #3121
• [Backport 2.6] enhance: improve AsyncMilvusClient connection error message (#3102) by @pymilvus-bot in #3122
• [Backport 2.6] fix: use deepcopy for struct_schema in add_field to prevent reuse issues (#3105) by @pymilvus-bot in #3123
• [Backport 2.6] [enhance]: Add some missing features to AsyncMilvusClient and refactor common code (#3108) by @pymilvus-bot in #3124
• [Backport 2.6] fix: error when function or function score fetch empty params (#3119) by @pymilvus-bot in #3125
• feat: suppot search with highlighter in pymilvus (#3110) by @aoiasd in #3127
• [cherry-pick]fix: AsyncMilvusClient.search support EmbeddingList (#3101) by @jac0626 in #3130

New Contributors

• @pymilvus-bot made their first contribution in #3121

Full Changelog: v2.6.4...v2.6.5

PyMilvus v2.5.18 Release Notes

What's Changed

• [cherry-pick] Breaking change: Rename Stage to Volume by @lentitude2tk in #3111

Full Changelog: v2.5.17...v2.5.18

PyMilvus v2.6.4 Release Notes

Highlights

1. Support for TimestampTZ Data Type (Introduced in Milvus v2.6.6)

PyMilvus now supports the TIMESTAMPTZ data type, allowing you to store and query timestamps with time zone information.

from pymilvus import MilvusClient, DataType

client = MilvusClient(uri="http://localhost:19530")

# 1. Define Schema with TimestampTZ
schema = MilvusClient.create_schema()
schema.add_field("id", DataType.INT64, is_primary=True)
schema.add_field("timestamp", DataType.TIMESTAMPTZ)
schema.add_field("vector", DataType.FLOAT_VECTOR, dim=128)

client.create_collection(collection_name="ts_collection", schema=schema)

# 2. Insert Data (using ISO 8601 format strings with timezone)
data = [
    {"id": 1, "timestamp": "2025-01-01T12:00:00+00:00", "vector": [0.1] * 128},
    {"id": 2, "timestamp": "2025-01-02T08:30:00-05:00", "vector": [0.2] * 128}
]
client.insert(collection_name="ts_collection", data=data)

# 3. Query Data
# Query records by ID to retrieve timestamps
res = client.query(
    collection_name="ts_collection",
    filter="id in [1, 2]",
    output_fields=["id", "timestamp"]
)
print(res)

Change Log

Features & Enhancements

• [TimestampTZ] Added support for TIMESTAMPTZ data type. (#3092)
• [BulkWriter] Supported STRUCT field data type in BulkWriter. (#3068)
• [BulkWriter] Supported Geometry data type in BulkWriter. (#3052)
• [BulkWriter] Added support for specifying a local temporary path in RemoteBulkWriter. (#3077)
• [STRUCT] Supported TypeParams definition in STRUCT fields. (#3055)
• [MilvusClient] Migrated 6 APIs to MilvusClient: flush_all, get_flush_all_state, list_loaded_segments, list_persistent_segments, get_server_type, and get_compaction_plans. (#3096)
• [Array] Supported numpy ndarray input for ARRAY fields in MilvusClient. (#3096)

Bug Fixes

• [Import] Added db_name parameter to list_import_jobs() to correctly list jobs under a specific database. (#3088)
• [Connection] Fixed a bug related to connection reset handling. (#3085)
• [Connection] Fixed an issue with connection caching when retrieving collection schema. (#3072)
• [STRUCT] Fixed a bug where querying STRUCT fields could incorrectly return an empty array. (#3074)
• [MilvusClient] Fixed an issue where insert() did not pass **kwargs (e.g., timeout) to the underlying implementation. (#3096)
• [Async] Fixed connection reuse issue by including event loop ID in connection alias. (#3096)
• [Async] Fixed async flush() not waiting for segments to be flushed. (#3090)
• [MilvusClient] Fixed key names in segment info retrieval. (#3096)
• [Error Handling] Added detailed traceback to error handler for better debugging. (#3096)

Full Changelog: v2.6.3...v2.6.4

PyMilvus v2.5.17 Release Notes

What's Changed

• enhance: allow user insert pk when autoid is enabled by @sunby in #3006
• Support insert serialized json (#3024) by @sunby in #3032
• fix: Bulk writer support geometry by @xiaocai2333 in #3053
• GCP supports using stage by @lentitude2tk in #3063

Full Changelog: v2.5.16...v2.5.17

PyMilvus v2.6.3 Release Notes

Highlights

1. Support for Array of Structs

PyMilvus now supports array of structs data types, allowing you to store and query complex nested data structures.

from pymilvus import MilvusClient, DataType

client = MilvusClient(uri="http://localhost:19530")

schema = client.create_schema(auto_id=False)
schema.add_field(field_name="id", datatype=DataType.INT64, is_primary=True)
schema.add_field(field_name="vector", datatype=DataType.FLOAT_VECTOR, dim=128)

struct_schema = MilvusClient.create_struct_field_schema()
struct_schema.add_field(field_name="name", datatype=DataType.VARCHAR, max_length=100)
struct_schema.add_field(field_name="age", datatype=DataType.INT64)
struct_schema.add_field(field_name="text", datatype=DataType.VARCHAR, max_length=65535)
struct_schema.add_field(field_name="emb", datatype=DataType.FLOAT_VECTOR, dim=128)
schema.add_field(
    field_name="user_info",
    datatype=DataType.ARRAY,
    element_type=DataType.STRUCT,
    struct_schema=struct_schema,
    max_capacity=1000,
)

client.create_collection(collection_name="users", schema=schema)

data = {
    "id": 1,
    "user_info": [
        {"name": "Alice", "age": 30, "text": "this is alice", "emb": [0.1] * 128},
        {"name": "Bob", "age": 20, "text": "this is bob", "emb": [0.2] * 128},
    ]
    "vector": [0.1] * 128
}

client.insert(collection_name="users", data=[data])

📖 Documentation Array of Structs, Compatible with Milvus 2.6.4+

2. Geometry Data Type Support

Query and search with geographic data using the new Geometry data type.

from pymilvus import MilvusClient, DataType

client = MilvusClient(uri="http://localhost:19530")

schema = MilvusClient.create_schema()
schema.add_field(field_name="id", datatype=DataType.INT64, is_primary=True)
schema.add_field(field_name="location", datatype=DataType.GEOMETRY, nullable=True)
schema.add_field(field_name="vector", datatype=DataType.FLOAT_VECTOR, dim=128)

client.create_collection(collection_name="geo_coll", schema=schema)

data = [
    {"id": 1, "location": "POINT(116.404 39.915)", "vector": [0.1] * 128},
    {"id": 2, "location": "POINT(-73.935 40.730)", "vector": [0.2] * 128}
]
client.insert(collection_name="locations", data=data)

results = client.query(
    collection_name="locations",
    filter=f'st_dwithin(location, "POINT(116.4 39.9)", 1000.0)',
    output_fields=["id", "location"]
)

📖 Documentation Geometry field, Compatible with Milvus 2.6.4+

3. Insert Primary Key with Auto ID Enabled

You can now insert custom primary keys even when auto_id is enabled, providing more flexibility in data management.

from pymilvus import MilvusClient, DataType

client = MilvusClient(uri="http://localhost:19530")

schema = MilvusClient.create_schema(auto_id=True)
schema.add_field(field_name="id", datatype=DataType.INT64, is_primary=True, auto_id=True)
schema.add_field(field_name="vector", datatype=DataType.FLOAT_VECTOR, dim=128)

client.create_collection(collection_name="my_collection", schema=schema, properties={"allow_insert_auto_id": "true"})

data = [
    {"id": 100, "vector": [0.1] * 128},
    {"id": 200, "vector": [0.2] * 128}
]
client.insert(collection_name="my_collection", data=data)

📖 Documentation Insert when AutoID is true, Compatible with Milvus 2.6.3+

4. L0 Compaction Support

Trigger L0 compaction manually to optimize storage and query performance.

from pymilvus import MilvusClient

client = MilvusClient(uri="http://localhost:19530")

client.compact(collection_name="my_collection", is_l0=True)

📖 Documentation Compact L0 segments, Compatible with Milvus 2.6.3+

5. ⚠️ Breaking Change: JSON Serialization for Strings

String values are now treated as JSON-serialized bytes. This is incompatible with previous versions.

Before (2.6.2 and earlier):

client.insert(collection_name="my_collection", data=[{"id": 1, "json_field": "a"}])

Now (2.6.3+):

import json

client.insert(collection_name="my_collection", data=[{"id": 1, "json_field": json.dumps("a")}])

Complete Change Log

Features

• Struct Support: SDK implementation for struct data type with field-level mmap configuration (#3026, #3037)
• Geo Data Type: Support for geographic data in insert, query, and search operations (#3043)
• L0 Compaction: Force L0 compact support (#3011)
• Function Scorer Ranker: Support using function scorer as ranker (#3010)
• Replicate Configuration: UpdateReplicateConfiguration API support (#3003)

Enhancements

• Async Client Improvements: Added retry mechanism and schema cache for AsyncMilvusClient (#3023)
• Allow user to insert primary key when auto_id is enabled (#3007)
• Update get_cost_from_status implementation (#3000)
• Enhanced remote storage usage result info (#3014)
• Remove annoying logs in MilvusClient (#3016)
• Raise error when ranker has unknown type (#3021)
• Support insert serialized JSON (#3031)
• Cherry pick multiple PRs from master branch (#3048)

Bug Fixes

• Fix hybrid_search to support EmbeddingList in request data (#3028)

Documentation

• Add apiKey note for bulk_import (#2997)

Contributors

Special thanks to all contributors who made this release possible:
@SpadeA-Tang, @XuanYang-cn, @aoiasd, @bigsheeper, @chasingegg, @lentitude2tk, @MrPresent-Han, @sunby, @xiaocai2333, @zhuwenxing

---

For more information, visit the PyMilvus GitHub repository.
Full Changelog: v2.6.2...v2.6.3

PyMilvus v2.6.2 Release Notes

What's Changed

• fix: Refine partial upsert to preserve unspecified fields (#2955) by @weiliu1031 in #2970
• fix: Correct validation logic for partial upsert requests (#2974) by @weiliu1031 in #2977
• feat: support group by json field(#2884) by @MrPresent-Han in #2980
• enhance: [2.6]Run actions when push to repo to enable cov compare by @XuanYang-cn in #2969
• support connectType if use oss-bucket by @lentitude2tk in #2988
• enhance: Sync PyMilvus milvus-proto commit to v2.6.2 by @XuanYang-cn in #2995

Full Changelog: v2.6.1...v2.6.2

PyMilvus v2.5.16 Release Notes

What's Changed

• support geo data type for insert,query and search for pymilvus 2.5 by @Yinwei-Yu in #2943
• fix: [25]Fix the existing version fmt by @XuanYang-cn in #2959
• fix: [2.5]Passing unknown req.is_refresh to wait by @XuanYang-cn in #2965
• [cherry-pick] support connectType if use oss-bucket by @lentitude2tk in #2989
• [cherry-pick] bulk_import add apiKey note by @lentitude2tk in #2998

Full Changelog: v2.5.15...v2.5.16

PyMilvus v2.6.1 Release Notes

What's Changed

• Avoid describe_collection when query by ids by @yhmo in #2930
• bulkImport add objectUrls/token paramster & add example use by @lentitude2tk in #2934
• support stageManager & stageFileManager by @lentitude2tk in #2935
• fix: Fix the existing version fmt by @XuanYang-cn in #2960
• enhance: Add unixmsec in every RPC call by @XuanYang-cn in #2961
• enhance: Multiple cherry picks from master branch by @XuanYang-cn in #2962
• fix: Passing unknown req.is_refresh to wait by @XuanYang-cn in #2964

Full Changelog: v2.6.0...v2.6.1

PyMilvus v2.5.15 Release Notes

What's Changed

• Compatible with the default behavior of free on the cloud by @lentitude2tk in #2912
• fix: Return new pk value for upsert when autoid=true by @yhmo in #2915
• fix: Multiple cherry-picks from master by @XuanYang-cn in #2906
• bulkImport add objectUrls/token paramster & add example use by @lentitude2tk in #2932
• support stageManager & stageFileManager by @lentitude2tk in #2920

Full Changelog: v2.5.14...v2.5.15