### What problem does this PR solve?
Migrate OpenAI-compatible chats to LiteLLM.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### Related issues
#10078
### What problem does this PR solve?
Integrate CometAPI provider.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
- [x] Documentation Update
### What problem does this PR solve?
Add LongCat-Flash-Chat from Meituan, deepseek v3.1 from SiliconFlow,
kimi-k2-09-05-preview and kimi-k2-turbo-preview from Moonshot.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
The total token was incorrectly accumulated when using the
OpenAI-API-Compatible api.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
Add exponential back-off for Chat LiteLLM. #9858.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
This revision performed a comprehensive check on LightRAG to ensure the
correctness of its implementation. It **did not involve** Entity
Resolution and Community Reports Generation. There is an example using
default entity types and the General chunking method, which shows good
results in both time and effectiveness. Moreover, response caching is
enabled for resuming failed tasks.
[The-Necklace.pdf](https://github.com/user-attachments/files/22042432/The-Necklace.pdf)
After:
```bash
Begin at:
Fri, 29 Aug 2025 16:48:03 GMT
Duration:
222.31 s
Progress:
16:48:04 Task has been received.
16:48:06 Page(1~7): Start to parse.
16:48:06 Page(1~7): OCR started
16:48:08 Page(1~7): OCR finished (1.89s)
16:48:11 Page(1~7): Layout analysis (3.72s)
16:48:11 Page(1~7): Table analysis (0.00s)
16:48:11 Page(1~7): Text merged (0.00s)
16:48:11 Page(1~7): Finish parsing.
16:48:12 Page(1~7): Generate 7 chunks
16:48:12 Page(1~7): Embedding chunks (0.29s)
16:48:12 Page(1~7): Indexing done (0.04s). Task done (7.84s)
16:48:17 Start processing for f421fb06849e11f0bdd32724b93a52b2: She had no dresses, no je...
16:48:17 Start processing for f421fb06849e11f0bdd32724b93a52b2: Her husband, already half...
16:48:17 Start processing for f421fb06849e11f0bdd32724b93a52b2: And this life lasted ten ...
16:48:17 Start processing for f421fb06849e11f0bdd32724b93a52b2: Then she asked, hesitatin...
16:49:30 Completed processing for f421fb06849e11f0bdd32724b93a52b2: She had no dresses, no je... after 1 gleanings, 21985 tokens.
16:49:30 Entities extraction of chunk 3 1/7 done, 12 nodes, 13 edges, 21985 tokens.
16:49:40 Completed processing for f421fb06849e11f0bdd32724b93a52b2: Finally, she replied, hes... after 1 gleanings, 22584 tokens.
16:49:40 Entities extraction of chunk 5 2/7 done, 19 nodes, 19 edges, 22584 tokens.
16:50:02 Completed processing for f421fb06849e11f0bdd32724b93a52b2: Then she asked, hesitatin... after 1 gleanings, 24610 tokens.
16:50:02 Entities extraction of chunk 0 3/7 done, 16 nodes, 28 edges, 24610 tokens.
16:50:03 Completed processing for f421fb06849e11f0bdd32724b93a52b2: And this life lasted ten ... after 1 gleanings, 24031 tokens.
16:50:04 Entities extraction of chunk 1 4/7 done, 24 nodes, 22 edges, 24031 tokens.
16:50:14 Completed processing for f421fb06849e11f0bdd32724b93a52b2: So they begged the jewell... after 1 gleanings, 24635 tokens.
16:50:14 Entities extraction of chunk 6 5/7 done, 27 nodes, 26 edges, 24635 tokens.
16:50:29 Completed processing for f421fb06849e11f0bdd32724b93a52b2: Her husband, already half... after 1 gleanings, 25758 tokens.
16:50:29 Entities extraction of chunk 2 6/7 done, 25 nodes, 35 edges, 25758 tokens.
16:51:35 Completed processing for f421fb06849e11f0bdd32724b93a52b2: The Necklace By Guy de Ma... after 1 gleanings, 27491 tokens.
16:51:35 Entities extraction of chunk 4 7/7 done, 39 nodes, 37 edges, 27491 tokens.
16:51:35 Entities and relationships extraction done, 147 nodes, 177 edges, 171094 tokens, 198.58s.
16:51:35 Entities merging done, 0.01s.
16:51:35 Relationships merging done, 0.01s.
16:51:35 ignored 7 relations due to missing entities.
16:51:35 generated subgraph for doc f421fb06849e11f0bdd32724b93a52b2 in 198.68 seconds.
16:51:35 run_graphrag f421fb06849e11f0bdd32724b93a52b2 graphrag_task_lock acquired
16:51:35 set_graph removed 0 nodes and 0 edges from index in 0.00s.
16:51:35 Get embedding of nodes: 9/147
16:51:35 Get embedding of nodes: 109/147
16:51:37 Get embedding of edges: 9/170
16:51:37 Get embedding of edges: 109/170
16:51:40 set_graph converted graph change to 319 chunks in 4.21s.
16:51:40 Insert chunks: 4/319
16:51:40 Insert chunks: 104/319
16:51:40 Insert chunks: 204/319
16:51:40 Insert chunks: 304/319
16:51:40 set_graph added/updated 147 nodes and 170 edges from index in 0.53s.
16:51:40 merging subgraph for doc f421fb06849e11f0bdd32724b93a52b2 into the global graph done in 4.79 seconds.
16:51:40 Knowledge Graph done (204.29s)
```
Before:
```bash
Begin at:
Fri, 29 Aug 2025 17:00:47 GMT
processDuration:
173.38 s
Progress:
17:00:49 Task has been received.
17:00:51 Page(1~7): Start to parse.
17:00:51 Page(1~7): OCR started
17:00:53 Page(1~7): OCR finished (1.82s)
17:00:57 Page(1~7): Layout analysis (3.64s)
17:00:57 Page(1~7): Table analysis (0.00s)
17:00:57 Page(1~7): Text merged (0.00s)
17:00:57 Page(1~7): Finish parsing.
17:00:57 Page(1~7): Generate 7 chunks
17:00:57 Page(1~7): Embedding chunks (0.31s)
17:00:57 Page(1~7): Indexing done (0.03s). Task done (7.88s)
17:00:57 created task graphrag
17:01:00 Task has been received.
17:02:17 Entities extraction of chunk 1 1/7 done, 9 nodes, 9 edges, 10654 tokens.
17:02:31 Entities extraction of chunk 2 2/7 done, 12 nodes, 13 edges, 11066 tokens.
17:02:33 Entities extraction of chunk 4 3/7 done, 9 nodes, 10 edges, 10433 tokens.
17:02:42 Entities extraction of chunk 5 4/7 done, 11 nodes, 14 edges, 11290 tokens.
17:02:52 Entities extraction of chunk 6 5/7 done, 13 nodes, 15 edges, 11039 tokens.
17:02:55 Entities extraction of chunk 3 6/7 done, 14 nodes, 13 edges, 11466 tokens.
17:03:32 Entities extraction of chunk 0 7/7 done, 19 nodes, 18 edges, 13107 tokens.
17:03:32 Entities and relationships extraction done, 71 nodes, 89 edges, 79055 tokens, 149.66s.
17:03:32 Entities merging done, 0.01s.
17:03:32 Relationships merging done, 0.01s.
17:03:32 ignored 1 relations due to missing entities.
17:03:32 generated subgraph for doc b1d9d3b6848711f0aacd7ddc0714c4d3 in 149.69 seconds.
17:03:32 run_graphrag b1d9d3b6848711f0aacd7ddc0714c4d3 graphrag_task_lock acquired
17:03:32 set_graph removed 0 nodes and 0 edges from index in 0.00s.
17:03:32 Get embedding of nodes: 9/71
17:03:33 Get embedding of edges: 9/88
17:03:34 set_graph converted graph change to 161 chunks in 2.27s.
17:03:34 Insert chunks: 4/161
17:03:34 Insert chunks: 104/161
17:03:34 set_graph added/updated 71 nodes and 88 edges from index in 0.28s.
17:03:34 merging subgraph for doc b1d9d3b6848711f0aacd7ddc0714c4d3 into the global graph done in 2.60 seconds.
17:03:34 Knowledge Graph done (153.18s)
```
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
- [x] Refactoring
- [x] Performance Improvement
### What problem does this PR solve?
Fix Ollama chat can only access localhost instance. #9806.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
All models pass the mock response tests, which means that if a model can
return the correct response, everything should work as expected.
However, not all models have been fully tested in a real environment,
the real API_KEY. I suggest actively monitoring the refactored models
over the coming period to ensure they work correctly and fixing them
step by step, or waiting to merge until most have been tested in
practical environment.
### Type of change
- [x] Refactoring
Updated constructors for base and derived classes in chat, embedding,
rerank, sequence2txt, and tts models to accept **kwargs. This change
improves extensibility and allows passing additional parameters without
breaking existing interfaces.
- [x] Bug Fix (non-breaking change which fixes an issue)
---------
Co-authored-by: IT: Sop.Son <sop.son@feavn.local>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
### What problem does this PR solve?
https://github.com/infiniflow/ragflow/issues/9177
The reason should be due to the gemin internal use a different parameter
name
`
max_output_tokens (int):
Optional. The maximum number of tokens to include in a
response candidate.
Note: The default value varies by model, see the
``Model.output_token_limit`` attribute of the ``Model``
returned from the ``getModel`` function.
This field is a member of `oneof`_ ``_max_output_tokens``.
`
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
#9082#6365
<u> **WARNING: it's not compatible with the older version of `Agent`
module, which means that `Agent` from older versions can not work
anymore.**</u>
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
fix error 429 api rate limit when building knowledge graph for all chat
model and Mistral embedding model.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
fix issue with `keep_alive=-1` for ollama chat model by allowing a user
to set an additional configuration option. It is no-breaking change
because it still uses a previous default value such as: `keep_alive=-1`
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
- [X] Performance Improvement
- [X] Other (please describe):
- Additional configuration option has been added to control behavior of
RAGFlow while working with ollama LLM
### What problem does this PR solve?
Add model provider DeepInfra. This model list comes from our community.
NOTE: most endpoints haven't been tested, but they should work as OpenAI
does.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
Correct the logging message from "OpenAI cat_with_tools" to "OpenAI
chat_with_tools" in the `_exceptions` method of the `Base` class to
accurately reflect the method name and improve error traceability.
### Type of change
- [x] Typo
### What problem does this PR solve?
Add xAI provider (experimental feature, requires user feedback).
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
This PR addresses an incompatibility issue with the Google Chat API by
correcting the message content format in the `GoogleChat` class.
Previously, the content was directly assigned to the "parts" field,
which did not align with the API's expected format. This change ensures
that messages are properly formatted with a "text" key within a
dictionary, as required by the API.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
[https://github.com/infiniflow/ragflow/issues/8324](url)
docker image version: v0.19.1
The `_clean_conf` function was not implemented in the `_chat` and
`chat_streamly` methods of the `GeminiChat` class, causing the error
"Unknown field for GenerationConfig: max_tokens" when the default LLM
config includes the "max_tokens" parameter.
**Buggy Code(ragflow/rag/llm/chat_model.py)**
```python
class GeminiChat(Base):
def __init__(self, key, model_name, base_url=None, **kwargs):
super().__init__(key, model_name, base_url=base_url, **kwargs)
from google.generativeai import GenerativeModel, client
client.configure(api_key=key)
_client = client.get_default_generative_client()
self.model_name = "models/" + model_name
self.model = GenerativeModel(model_name=self.model_name)
self.model._client = _client
def _clean_conf(self, gen_conf):
for k in list(gen_conf.keys()):
if k not in ["temperature", "top_p"]:
del gen_conf[k]
return gen_conf
def _chat(self, history, gen_conf):
from google.generativeai.types import content_types
system = history[0]["content"] if history and history[0]["role"] == "system" else ""
hist = []
for item in history:
if item["role"] == "system":
continue
hist.append(deepcopy(item))
item = hist[-1]
if "role" in item and item["role"] == "assistant":
item["role"] = "model"
if "role" in item and item["role"] == "system":
item["role"] = "user"
if "content" in item:
item["parts"] = item.pop("content")
if system:
self.model._system_instruction = content_types.to_content(system)
response = self.model.generate_content(hist, generation_config=gen_conf)
ans = response.text
return ans, response.usage_metadata.total_token_count
def chat_streamly(self, system, history, gen_conf):
from google.generativeai.types import content_types
if system:
self.model._system_instruction = content_types.to_content(system)
#❌_clean_conf was not implemented
for k in list(gen_conf.keys()):
if k not in ["temperature", "top_p", "max_tokens"]:
del gen_conf[k]
for item in history:
if "role" in item and item["role"] == "assistant":
item["role"] = "model"
if "content" in item:
item["parts"] = item.pop("content")
ans = ""
try:
response = self.model.generate_content(history, generation_config=gen_conf, stream=True)
for resp in response:
ans = resp.text
yield ans
yield response._chunks[-1].usage_metadata.total_token_count
except Exception as e:
yield ans + "\n**ERROR**: " + str(e)
yield 0
```
**Implement the _clean_conf function**
```python
class GeminiChat(Base):
def __init__(self, key, model_name, base_url=None, **kwargs):
super().__init__(key, model_name, base_url=base_url, **kwargs)
from google.generativeai import GenerativeModel, client
client.configure(api_key=key)
_client = client.get_default_generative_client()
self.model_name = "models/" + model_name
self.model = GenerativeModel(model_name=self.model_name)
self.model._client = _client
def _clean_conf(self, gen_conf):
for k in list(gen_conf.keys()):
if k not in ["temperature", "top_p"]:
del gen_conf[k]
return gen_conf
def _chat(self, history, gen_conf):
from google.generativeai.types import content_types
#✅ implement _clean_conf to remove the wrong parameters
gen_conf = self._clean_conf(gen_conf)
system = history[0]["content"] if history and history[0]["role"] == "system" else ""
hist = []
for item in history:
if item["role"] == "system":
continue
hist.append(deepcopy(item))
item = hist[-1]
if "role" in item and item["role"] == "assistant":
item["role"] = "model"
if "role" in item and item["role"] == "system":
item["role"] = "user"
if "content" in item:
item["parts"] = item.pop("content")
if system:
self.model._system_instruction = content_types.to_content(system)
response = self.model.generate_content(hist, generation_config=gen_conf)
ans = response.text
return ans, response.usage_metadata.total_token_count
def chat_streamly(self, system, history, gen_conf):
from google.generativeai.types import content_types
#✅ implement _clean_conf to remove the wrong parameters
gen_conf = self._clean_conf(gen_conf)
if system:
self.model._system_instruction = content_types.to_content(system)
#✅Removed duplicate parameter filtering logic "for k in list(gen_conf.keys()):"
for item in history:
if "role" in item and item["role"] == "assistant":
item["role"] = "model"
if "content" in item:
item["parts"] = item.pop("content")
ans = ""
try:
response = self.model.generate_content(history, generation_config=gen_conf, stream=True)
for resp in response:
ans = resp.text
yield ans
yield response._chunks[-1].usage_metadata.total_token_count
except Exception as e:
yield ans + "\n**ERROR**: " + str(e)
yield 0
```
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
---------
Co-authored-by: Kevin Hu <kevinhu.sh@gmail.com>
### What problem does this PR solve?
This is a cherry-pick from #7781 as requested.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
Co-authored-by: Kevin Hu <kevinhu.sh@gmail.com>
### What problem does this PR solve?
- Simplify AzureChat constructor by passing base_url directly
- Clean up spacing and formatting in chat_model.py
- Remove redundant parentheses and improve code consistency
- #8423
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
Fix auto-keyword and auto-question fail with qwq model. #8189
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
Hello, our use case requires LLM agent to invoke some tools, so I made a
simple implementation here.
This PR does two things:
1. A simple plugin mechanism based on `pluginlib`:
This mechanism lives in the `plugin` directory. It will only load
plugins from `plugin/embedded_plugins` for now.
A sample plugin `bad_calculator.py` is placed in
`plugin/embedded_plugins/llm_tools`, it accepts two numbers `a` and `b`,
then give a wrong result `a + b + 100`.
In the future, it can load plugins from external location with little
code change.
Plugins are divided into different types. The only plugin type supported
in this PR is `llm_tools`, which must implement the `LLMToolPlugin`
class in the `plugin/llm_tool_plugin.py`.
More plugin types can be added in the future.
2. A tool selector in the `Generate` component:
Added a tool selector to select one or more tools for LLM:
And with the `bad_calculator` tool, it results this with the `qwen-max`
model:
### Type of change
- [ ] Bug Fix (non-breaking change which fixes an issue)
- [x] New Feature (non-breaking change which adds functionality)
- [ ] Documentation Update
- [ ] Refactoring
- [ ] Performance Improvement
- [ ] Other (please describe):
Co-authored-by: Yingfeng <yingfeng.zhang@gmail.com>
### What problem does this PR solve?
Fix: local variable referenced before assignment. #6803
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
This PR introduces **primitive support for function calls**,
enabling the system to handle basic function call capabilities.
However, this feature is currently experimental and **not yet enabled
for general use**, as it is only supported by a subset of models,
namely, Qwen and OpenAI models.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
https://github.com/infiniflow/ragflow/issues/6138
This PR is going to support vision llm for gpustack, modify url path
from `/v1-openai` to `/v1`
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
# Dynamic Context Window Size for Ollama Chat
## Problem Statement
Previously, the Ollama chat implementation used a fixed context window
size of 32768 tokens. This caused two main issues:
1. Performance degradation due to unnecessarily large context windows
for small conversations
2. Potential business logic failures when using smaller fixed sizes
(e.g., 2048 tokens)
## Solution
Implemented a dynamic context window size calculation that:
1. Uses a base context size of 8192 tokens
2. Applies a 1.2x buffer ratio to the total token count
3. Adds multiples of 8192 tokens based on the buffered token count
4. Implements a smart context size update strategy
## Implementation Details
### Token Counting Logic
```python
def count_tokens(text):
"""Calculate token count for text"""
# Simple calculation: 1 token per ASCII character
# 2 tokens for non-ASCII characters (Chinese, Japanese, Korean, etc.)
total = 0
for char in text:
if ord(char) < 128: # ASCII characters
total += 1
else: # Non-ASCII characters
total += 2
return total
```
### Dynamic Context Calculation
```python
def _calculate_dynamic_ctx(self, history):
"""Calculate dynamic context window size"""
# Calculate total tokens for all messages
total_tokens = 0
for message in history:
content = message.get("content", "")
content_tokens = count_tokens(content)
role_tokens = 4 # Role marker token overhead
total_tokens += content_tokens + role_tokens
# Apply 1.2x buffer ratio
total_tokens_with_buffer = int(total_tokens * 1.2)
# Calculate context size in multiples of 8192
if total_tokens_with_buffer <= 8192:
ctx_size = 8192
else:
ctx_multiplier = (total_tokens_with_buffer // 8192) + 1
ctx_size = ctx_multiplier * 8192
return ctx_size
```
### Integration in Chat Method
```python
def chat(self, system, history, gen_conf):
if system:
history.insert(0, {"role": "system", "content": system})
if "max_tokens" in gen_conf:
del gen_conf["max_tokens"]
try:
# Calculate new context size
new_ctx_size = self._calculate_dynamic_ctx(history)
# Prepare options with context size
options = {
"num_ctx": new_ctx_size
}
# Add other generation options
if "temperature" in gen_conf:
options["temperature"] = gen_conf["temperature"]
if "max_tokens" in gen_conf:
options["num_predict"] = gen_conf["max_tokens"]
if "top_p" in gen_conf:
options["top_p"] = gen_conf["top_p"]
if "presence_penalty" in gen_conf:
options["presence_penalty"] = gen_conf["presence_penalty"]
if "frequency_penalty" in gen_conf:
options["frequency_penalty"] = gen_conf["frequency_penalty"]
# Make API call with dynamic context size
response = self.client.chat(
model=self.model_name,
messages=history,
options=options,
keep_alive=60
)
return response["message"]["content"].strip(), response.get("eval_count", 0) + response.get("prompt_eval_count", 0)
except Exception as e:
return "**ERROR**: " + str(e), 0
```
## Benefits
1. **Improved Performance**: Uses appropriate context windows based on
conversation length
2. **Better Resource Utilization**: Context window size scales with
content
3. **Maintained Compatibility**: Works with existing business logic
4. **Predictable Scaling**: Context growth in 8192-token increments
5. **Smart Updates**: Context size updates are optimized to reduce
unnecessary model reloads
## Future Considerations
1. Fine-tune buffer ratio based on usage patterns
2. Add monitoring for context window utilization
3. Consider language-specific token counting optimizations
4. Implement adaptive threshold based on conversation patterns
5. Add metrics for context size update frequency
---------
Co-authored-by: Kevin Hu <kevinhu.sh@gmail.com>
