my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

#1	# Migration Guide: Upgrading to mem0 1.0.0
#2
#3	## TL;DR
#4
#5	What changed? We simplified the API by removing confusing version parameters. Now everything returns a consistent format: `{"results": [...]}`.
#6
#7	What you need to do:
#8	1. Upgrade: `pip install mem0ai==1.0.0`
#9	2. Remove `version` and `output_format` parameters from your code
#10	3. Update response handling to use `result["results"]` instead of treating responses as lists
#11
#12	Time needed: ~5-10 minutes for most projects
#13
#14	---
#15
#16	## Quick Migration Guide
#17
#18	### 1. Install the Update
#19
#20	```bash
#21	pip install mem0ai==1.0.0
#22	```
#23
#24	### 2. Update Your Code
#25
#26	If you're using the Memory API:
#27
#28	```python
#29	# Before
#30	memory = Memory(config=MemoryConfig(version="v1.1"))
#31	result = memory.add("I like pizza")
#32
#33	# After
#34	memory = Memory() # That's it - version is automatic now
#35	result = memory.add("I like pizza")
#36	```
#37
#38	If you're using the Client API:
#39
#40	```python
#41	# Before
#42	client.add(messages, output_format="v1.1")
#43	client.search(query, version="v2", output_format="v1.1")
#44
#45	# After
#46	client.add(messages) # Just remove those extra parameters
#47	client.search(query)
#48	```
#49
#50	### 3. Update How You Handle Responses
#51
#52	All responses now use the same format: a dictionary with `"results"` key.
#53
#54	```python
#55	# Before - you might have done this
#56	result = memory.add("I like pizza")
#57	for item in result: # Treating it as a list
#58	print(item)
#59
#60	# After - do this instead
#61	result = memory.add("I like pizza")
#62	for item in result["results"]: # Access the results key
#63	print(item)
#64
#65	# Graph relations (if you use them)
#66	if "relations" in result:
#67	for relation in result["relations"]:
#68	print(relation)
#69	```
#70
#71	---
#72
#73	## Enhanced Message Handling
#74
#75	The platform client (MemoryClient) now supports the same flexible message formats as the OSS version:
#76
#77	```python
#78	from mem0 import MemoryClient
#79
#80	client = MemoryClient(api_key="your-key")
#81
#82	# All three formats now work:
#83
#84	# 1. Single string (automatically converted to user message)
#85	client.add("I like pizza", user_id="alice")
#86
#87	# 2. Single message dictionary
#88	client.add({"role": "user", "content": "I like pizza"}, user_id="alice")
#89
#90	# 3. List of messages (conversation)
#91	client.add([
#92	{"role": "user", "content": "I like pizza"},
#93	{"role": "assistant", "content": "I'll remember that!"}
#94	], user_id="alice")
#95	```
#96
#97	### Async Mode Configuration
#98
#99	The `async_mode` parameter now defaults to `True` but can be configured:
#100
#101	```python
#102	# Default behavior (async_mode=True)
#103	client.add(messages, user_id="alice")
#104
#105	# Explicitly set async mode
#106	client.add(messages, user_id="alice", async_mode=True)
#107
#108	# Disable async mode if needed
#109	client.add(messages, user_id="alice", async_mode=False)
#110	```
#111
#112	Note: `async_mode=True` provides better performance for most use cases. Only set it to `False` if you have specific synchronous processing requirements.
#113
#114	---
#115
#116	## That's It!
#117
#118	For most users, that's all you need to know. The changes are:
#119	- ✅ No more `version` or `output_format` parameters
#120	- ✅ Consistent `{"results": [...]}` response format
#121	- ✅ Cleaner, simpler API
#122
#123	---
#124
#125	## Common Issues
#126
#127	Getting `KeyError: 'results'`?
#128
#129	Your code is still treating the response as a list. Update it:
#130	```python
#131	# Change this:
#132	for memory in response:
#133
#134	# To this:
#135	for memory in response["results"]:
#136	```
#137
#138	Getting `TypeError: unexpected keyword argument`?
#139
#140	You're still passing old parameters. Remove them:
#141	```python
#142	# Change this:
#143	client.add(messages, output_format="v1.1")
#144
#145	# To this:
#146	client.add(messages)
#147	```
#148
#149	Seeing deprecation warnings?
#150
#151	Remove any explicit `version="v1.0"` from your config:
#152	```python
#153	# Change this:
#154	memory = Memory(config=MemoryConfig(version="v1.0"))
#155
#156	# To this:
#157	memory = Memory()
#158	```
#159
#160	---
#161
#162	## What's New in 1.0.0
#163
#164	- Better vector stores: Fixed OpenSearch and improved reliability across all stores
#165	- Cleaner API: One way to do things, no more confusing options
#166	- Enhanced GCP support: Better Vertex AI configuration options
#167	- Flexible message input: Platform client now accepts strings, dicts, and lists (aligned with OSS)
#168	- Configurable async_mode: Now defaults to `True` but users can override if needed
#169
#170	---
#171
#172	## Need Help?
#173
#174	- Check [GitHub Issues](https://github.com/mem0ai/mem0/issues)
#175	- Read the [documentation](https://docs.mem0.ai/)
#176	- Open a new issue if you're stuck
#177
#178	---
#179
#180	## Advanced: Configuration Changes
#181
#182	If you configured vector stores with version:
#183
#184	```python
#185	# Before
#186	config = MemoryConfig(
#187	version="v1.1",
#188	vector_store=VectorStoreConfig(...)
#189	)
#190
#191	# After
#192	config = MemoryConfig(
#193	vector_store=VectorStoreConfig(...)
#194	)
#195	```
#196
#197	---
#198
#199	## Testing Your Migration
#200
#201	Quick sanity check:
#202
#203	```python
#204	from mem0 import Memory
#205
#206	memory = Memory()
#207
#208	# Add should return a dict with "results"
#209	result = memory.add("I like pizza", user_id="test")
#210	assert "results" in result
#211
#212	# Search should return a dict with "results"
#213	search = memory.search("food", user_id="test")
#214	assert "results" in search
#215
#216	# Get all should return a dict with "results"
#217	all_memories = memory.get_all(user_id="test")
#218	assert "results" in all_memories
#219
#220	print("✅ Migration successful!")
#221	```
#222

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public