### Related issues
#10078
### What problem does this PR solve?
Integrate DeerAPI provider.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
- [x] Documentation Update
Co-authored-by: DeerAPI <tensor.null@gmail.com>
### What problem does this PR solve?
Add support for international Dashscope service. #10340
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
Rename the CometEmbed and CometSeq2txt classes to CometAPIEmbed and
CometAPISeq2txt, and correct supported_models.mdx.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
Fix invalid COMPONENT_EXEC_TIMEOUT. #10273
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
issue:
[Bug]: ERROR: list index out of range #10188
change:
fix a potential list index out of range error in chat response parsing
by adding explicit checks for empty choices.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
Revert back to chat.completions.
### Type of change
- [ ] Bug Fix (non-breaking change which fixes an issue)
- [ ] New Feature (non-breaking change which adds functionality)
- [ ] Documentation Update
- [ ] Refactoring
- [ ] Performance Improvement
- [x] Other (please describe):
Revert back to chat.completions.
### What problem does this PR solve?
Currently, Azure OpenAI returns one minute Quota limit responses when
chat API is utilized. This change is needed in order to be able to
process almost any documents using models deployed in Azure Foundry.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
Fix broken imports
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
---------
Signed-off-by: jinhai <haijin.chn@gmail.com>
### What problem does this PR solve?
Fix: resolve hash collisions by switching to UUID &correct logic for
always-true statements, solved: #10165
Feat: Update GPT api integration, solved: #10204
Feat: Support qianwen-deepresearch, solved: #10163
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
issue:
[Bug]: anthropic model have not baseurl selecting,need add #8546
change:
This PR adds support for using Anthropic models through a third-party
API by allowing a custom base_url.
It ensures compatibility with both the official Anthropic endpoint and
external providers.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
fix the rerank_model condition logic by correcting the np.isclose check.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
Migrate OpenAI-compatible chats to LiteLLM.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
Related PR:
Feat: add CometAPI to LLMFactory and update related mappings #10119
Change:
Fixes the issue where the embedding model in CometAPI was not being
called correctly
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
- [x] New Feature (non-breaking change which adds functionality)
---------
Co-authored-by: TensorNull <tensor.null@gmail.com>
### 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?
Dataflow support audio. And fix giteeAI's sequence2text model.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
- [x] New Feature (non-breaking change which adds functionality)
### What problem does this PR solve?
fix text input exceed token num limit when using siliconflow's embedding
model BAAI/bge-large-zh-v1.5 and BAAI/bge-large-en-v1.5, truncate before
input.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
### What problem does this PR solve?
Dataflow supports Spreadsheet and Word processor document
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
### 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?
The default value for OpenAI '/v1/embeddings' parameter
'encoding_format' is 'base64'. Use 'float' explicitly to avoid base64
encoding & decoding, larger data size.
https://github.com/openai/openai-python/blob/main/src/openai/resources/embeddings.py
if not is_given(encoding_format):
params["encoding_format"] = "base64"
### Type of change
- [x] Performance Improvement
### 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?
for total_token_count method use if to check first, to improve the
performance when we need to handle exception cases
### Type of change
- [x] Refactoring
### What problem does this PR solve?
Refactor OpenAI to enable audio parsing.
### Type of change
- [x] Bug Fix (non-breaking change which fixes an issue)
- [x] Refactoring
### What problem does this PR solve?
Fix Gemini parameters error.
### 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?
1. Use input length to prepare res
2. Adjust torch_empty_cache code location
### Type of change
- [x] Refactoring
- [x] Performance Improvement
fix: preserve correct MIME & unify data URL handling for vision inputs
(relates #9248)
- Updated image2base64() to return a full data URL
(data:image/<fmt>;base64,...) with accurate MIME
- Removed hardcoded image/jpeg in Base._image_prompt(); pass through
data URLs and default raw base64 to image/png
- Set AnthropicCV._image_prompt() raw base64 media_type default to
image/png
- Ensures MIME type matches actual image content, fixing “cannot process
base64 image” errors on vLLM/OpenAI-compatible backends
### What problem does this PR solve?
This PR fixes a compatibility issue where base64-encoded images sent to
vision models (e.g., vLLM/OpenAI-compatible backends) were rejected due
to mismatched MIME type or incorrect decoding.
Previously, the backend:
- Always converted raw base64 into data:image/jpeg;base64,... even if
the actual content was PNG.
- In some cases, base64 decoding was attempted on the full data URL
string instead of the pure base64 part.
This caused errors like:
```
cannot process base64 image
failed to decode base64 string: illegal base64 data at input byte 0
```
by strict validators such as vLLM.
With this fix, the MIME type in the request now matches the actual image
content, and data URLs are correctly handled or passed through, ensuring
vision models can decode and process images reliably.
### 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)
