langgenius/dify 0.15.3 on GitHub

✨ What’s New in v0.15.3? ✨

Hey folks, it's time for another update, and we've got some exciting additions and improvements rolling out with version v0.15.3. Let’s dive into the highlights:

🚀 New Model Support

DeepSeek Models: DeepSeek R1 is now supported across several platforms, including SiliconFlow, Azure AI Foundry, Ollama, Volcengine, OpenRouter, Nvidia Catalog, and more. This opens up new possibilities for leveraging AI reasoning and insights across diverse environments.
Gemini 2.0 Series: Introducing the latest models in the Gemini 2.0 series, such as Gemini 2.0 Flash 001 and Gemini 2.0 Pro Exp. These models deliver exceptional performance and efficiency, perfect for high-precision, fast-processing tasks.

⚙️ Enhancements

Firecrawl Upgrade: Updated to the v1 API by @ftonato, providing improved performance and more refined results.
GLM-4-AIR-0111 Support: Dive into advanced natural language processing tasks more effectively with this addition by @Jhvcc.
Agent Thinking Content Display: Visualize agent thinking paths with DeepSeek’s R1 model, providing better understanding and debugging capabilities, courtesy of @hjlarry.

🛠️ Bug Fixes

SSRF Proxy File Descriptor Leak Fix: Eliminated file descriptor leaks to enhance performance in concurrent environments, thanks to @ZuzooVn.
Page Crash on String Variable Property Access: Resolved issues where improper string handling could cause unintentional crashes, fixed by @iamjoel.

Keep exploring and integrating these new capabilities into your workflows. As always, happy building, and we look forward to hearing how you leverage these updates in your projects! 🔍🚀

Upgrade Guide

Docker compose deployments

Warning

The files in the docker-legacy directory will soon stop being maintained and will be removed from the repository. If you are still using them, please switch to the new version as soon as possible.

Back up your customized docker-compose YAML file (optional)

cd docker
cp docker-compose.yaml docker-compose.yaml.$(date +%s).bak

Get the latest code from the main branch
```
git checkout main
git pull origin main
```
Stop the service，Command, please execute in the docker directory
```
docker compose down
```

Back up data

tar -cvf volumes-$(date +%s).tgz volumes

Upgrade services
```
docker compose up -d
```

Source Code deployments

Stop API server, Worker and Web frontend Server.
Get the latest code from the release branch:
```
git checkout 0.15.3
```
Update Python dependencies:
```
cd api
poetry install
```
Then, let's run the migration script:
```
poetry run flask db upgrade
```
Finally, run API server, Worker and Web frontend Server again.

What's Changed

feat: add pg vector index by @huangzhuo1949 in #12338
feat: add giteeAI risk control identification. by @jiandanfeng in #12946
fix: tools translate search by @Jhvcc in #12950
fix: get property of string type variable caused page crash by @iamjoel in #12969
refactor: Update Firecrawl to use v1 API by @ftonato in #12574
Feat/prompt editor dark theme by @zxhlyh in #12976
feat: add glm-4-air-0111 by @Jhvcc in #12997
fix: wrong zh_Hans translation: Ohio by @iwaim in #13006
Change typo in translation by @y10ab1 in #13004
chore: add Japanese translation: model_providers/bedrock by @iwaim in #13016
feat: add agent thinking content display of deepseek R1 by @hjlarry in #12949
support bedrock kb: retrieve and generate by @warren830 in #13027
fix: Unexpected tag creation when pressing enter during tag conversion by @miya in #13041
Update Novita AI predefined models by @jasonhp in #13045
fix: adjust iteration node dark style by @douxc in #13051
feat: add inner API to create workspace without requiring email by @zhangx1n in #13021
fix:Improve MIME type detection for remote URL uploads using python-magic by @yjc980121 in #12693
fix: "Model schema not found" error only in agents (#12655) by @jiandanfeng in #12760
fix qwen-vl blocking mode by @hieheihei in #13052
Set default LOG_LEVEL to INFO for celery workers and beat by @osaimi in #13066
Fixed Novita AI color and added DeepSeek R1 model by @jasonhp in #13074
fix: fix the formatter is not applied on log file by @acelyc111 in #12704
refactor: reduce duplciate code by inheritance by @acelyc111 in #13073
refactor: Update Firecrawl API parameters and default settings by @ftonato in #13082
add tongyi qwen2.5-14b/7b-instruct-1m model by @hieheihei in #13089
refactor: avoid to use extra space when finding model by name by @acelyc111 in #13043
feat: switch to chat messages before regenerated by @xuzuodong in #11301
fix: deduct LLM quota after processing invoke result by @laipz8200 in #13075
test(huggingface_hub): Skip the failed test temporarily. by @laipz8200 in #13142
chore: fix wrong VectorType match case by @tmokmss in #13123
feature. add feat to modify metadata via dataset api by @aplio in #13116
pass model and cost so that langfuse can show cost by @obadakhalili in #13117
feat: the think tag display of siliconflow's deepseek r1 by @hjlarry in #13153
chore: upgrade libldap2 by @ZuzooVn in #13158
nitpick: fix small typos in template.en.mdx by @aplio in #13156
add azure openai api version 2024-12-01-preview by @kei-yamazaki in #13135
fix: not install libmagic raise error by @hjlarry in #13146
fix: set indexing technique from dataset during update-by-text by @aplio in #13155
Solve circular dependency issue between workflow/constants.ts file and default.ts file by @obadakhalili in #13165
chore(pyproject): Add type stubs into pyproject.toml by @laipz8200 in #13145
Install apt dependencies using bookworm source, consistent with base image. Remove unnecessary, error-prone pins by @deephbz in #13176
fix: Dollar Sign Handling in Markdown by @Riddhimaan-Senapati in #13178
chore: add tests for build docker image when dockerfile changed by @bowenliang123 in #10732
Modify params for bedrock retrieve generate by @warren830 in #13182
feat: bedrock_endpoint_url by @2284730142 in #12838
Feat: add linkedin to readme by @Riddhimaan-Senapati in #13203
fix: SSRF proxy file descriptor leak in concurrent requests by @ZuzooVn in #13108
Use DefaultAzureCredential for managed identity in azure blob extention by @te-chan2 in #11559
fix: mypy checks violation in AzureBlobStorage by @bowenliang123 in #13215
chore(model_runtime): Move deepseek ahead in the providers list. by @laipz8200 in #13197
Fix new variables in the conversation opener would override prompt_variables by @HQidea in #13191
chore: squash system dependencies installation steps by @bowenliang123 in #13206
feat: Add o3-mini and o3-mini-2025-01-31 model variants by @Joshbly in #13129
feat(model_providers): Add Groq DeepSeek-R1-Distill-Llama-70b by @laipz8200 in #13229
feat: added deepseek r1 and v3 to siliconflow by @Riddhimaan-Senapati in #13238
fix: updated _position.yaml to include the latest model already integ… by @Riddhimaan-Senapati in #13245
feat: add think display for volcengine and generic openapi by @xhebox in #13234
feat: add parameter "reasoning_effort" and Openai o3-mini by @k-zaku in #13243
feat: added models :gemini 2.0 flash 001 and gemini 2.0 pro exp 02-05 by @Riddhimaan-Senapati in #13247
feat(model): add gemini-2.0-flash-thinking-exp-01-21 by @wangiii in #13230
fix: correct linewrap think display in generic openai api by @xhebox in #13260
feat(model): add gemini-2.0 model by @wangiii in #13266
feat(model_providers): Support deepseek for Azure AI Foundry by @laipz8200 in #13267
Fix bug large data no render by @weiwenyan-dev in #12683
feat(model_providers): Support deepseek-r1 for Nvidia Catalog by @laipz8200 in #13269
Upgrade oracle models by @engchina in #13174
chore(deps): bump katex from 0.16.10 to 0.16.21 in /web by @dependabot in #13270
fix: add dataset edit permissions by @WTW0313 in #13223
feat:add deepseek r1 think display for ollama provider by @leslie2046 in #13272
feat: add deepseek models for volcengine provider by @xhdd123321 in #13283
fix: incorrect transferMethod assignment for remote file by @le0zh in #13286
feat:add think tag display for xinference deepseek r1 by @leslie2046 in #13291
feat: added a couple of models not defined in vertex ai, that were already … by @Riddhimaan-Senapati in #13296
Update README_TR.md by @kemalmeler in #13294
feat(credits): Allow to configure model-credit mapping by @laipz8200 in #13274
feat(model): add deepseek-r1 for openrouter by @sinomoe in #13312
fix(credits): require model name equals by @laipz8200 in #13314
chore: refactor the OpenAICompatible and improve thinking display by @hjlarry in #13299
Fix the problem of Workflow terminates after parallel tasks execution, merge node not triggered by @lazyFrogLOL in #12498
chore(*): Update style of thinking by @laipz8200 in #13319
chore:improve thinking display for llm from xinference and ollama pro… by @leslie2046 in #13318
fix: Azure AI Foundry model cannot be used in the workflow by @laipz8200 in #13323
feat: add completion mode and context size options for LLM configuration by @laipz8200 in #13325
chore(*): Bump version to 0.15.3 by @laipz8200 in #13308

New Contributors

@Jhvcc made their first contribution in #12950
@ftonato made their first contribution in #12574
@y10ab1 made their first contribution in #13004
@zhangx1n made their first contribution in #13021
@yjc980121 made their first contribution in #12693
@hieheihei made their first contribution in #13052
@osaimi made their first contribution in #13066
@aplio made their first contribution in #13116
@obadakhalili made their first contribution in #13117
@deephbz made their first contribution in #13176
@Riddhimaan-Senapati made their first contribution in #13178
@2284730142 made their first contribution in #12838
@te-chan2 made their first contribution in #11559
@HQidea made their first contribution in #13191
@Joshbly made their first contribution in #13129
@xhebox made their first contribution in #13234
@weiwenyan-dev made their first contribution in #12683
@engchina made their first contribution in #13174
@kemalmeler made their first contribution in #13294
@lazyFrogLOL made their first contribution in #12498

Full Changelog: 0.15.2...0.15.3

langgenius/dify 0.15.3 v0.15.3 on GitHub