UN-2725 Updated Lllama-index version and related dependencies to Support GPT-5 models#196
Conversation
Summary by CodeRabbit
WalkthroughDependency versions updated across multiple LlamaIndex-related packages and embeddings/vector-store/LLM backends in pyproject.toml. SDK semantic version bumped from v0.76.1 to v0.77.0 in src/unstract/sdk/init.py. No code logic changes. Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Tip 🔌 Remote MCP (Model Context Protocol) integration is now available!Pro plan users can now connect to remote MCP servers from the Integrations page. Connect with popular remote MCPs such as Notion and Linear to add more context to your reviews and chats. ✨ Finishing Touches
🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. CodeRabbit Commands (Invoked using PR/Issue comments)Type Other keywords and placeholders
CodeRabbit Configuration File (
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (5)
src/unstract/sdk/__init__.py (1)
1-1: Confirm PEP 440 compliance of version string; consider dropping the leading "v".Some build backends enforce strict PEP 440. If
__version__ = "v0.77.0"flows into wheel metadata, publishing could fail or normalize unexpectedly. If CI has accepted the "v" historically, keep it; otherwise, prefer "0.77.0".Apply if you choose to align with PEP 440:
-__version__ = "v0.77.0" +__version__ = "0.77.0"pyproject.toml (4)
39-44: Heavy embedding stack in base install — consider moving to extras to reduce footprint.Installing all embeddings by default increases image size and cold-start times. Suggest an extra like
embeddings-alland keep base to OpenAI only.Proposed split:
-[project] -dependencies = [ - ... - "llama-index-embeddings-openai==0.5.0", - "llama-index-embeddings-ollama==0.8.1", - "llama-index-embeddings-bedrock==0.6.1", - "llama-index-embeddings-vertex==0.4.0", - ... -] +[project] +dependencies = [ + ... + "llama-index-embeddings-openai==0.5.0", + ... +] + +[project.optional-dependencies] +embeddings-all = [ + "llama-index-embeddings-ollama==0.8.1", + "llama-index-embeddings-bedrock==0.6.1", + "llama-index-embeddings-vertex==0.4.0", +]
47-50: Vector store plugins default-installed — consider extras to avoid unnecessary client SDKs.Milvus/Weaviate/Pinecone/Qdrant SDKs pull sizeable deps. Ship as extras (
vector-stores-allor per store) unless they’re required at runtime for every deployment.I can provide a minimal diff mirroring the embeddings split if you’d like.
51-61: OpenAI/Anthropic/Mistral backends: ensure transitive SDK versions are pinned/compatible for GPT‑5 and new APIs.
llama-index-llms-openai==0.5.4typically requiresopenai>=1.x. It’s pulled transitively; consider pinning a tested range (e.g.,openai>=1.40,<2) to avoid sudden breaking changes.- Same for
anthropic(transitive viallama-index-llms-anthropic), andpinecone-clientpulled by vector stores.If you want, I’ll add a lightweight constraints file to bound critical transitive deps.
75-79: Align optional extras and test deps to avoid drift.
gcsfs/s3fs/adlfsversions differ in operators between extras (~=) and tests (==). Minor, but pinning style drift can lead to subtle env diffs. Consider using the same operator across both.Example:
-test = [ - "parameterized==0.9.0", - "pytest==8.3.3", - "pytest-mock==3.14.0", - "gcsfs==2024.10.0", - "s3fs==2024.10.0", - "adlfs~=2024.7.0", -] +test = [ + "parameterized==0.9.0", + "pytest==8.3.3", + "pytest-mock==3.14.0", + "gcsfs~=2024.10.0", + "s3fs~=2024.10.0", + "adlfs~=2024.7.0", +]Also applies to: 91-98
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
Cache: Disabled due to Reviews > Disable Cache setting
Knowledge Base: Disabled due to Reviews > Disable Knowledge Base setting
⛔ Files ignored due to path filters (1)
uv.lockis excluded by!**/*.lock
📒 Files selected for processing (2)
pyproject.toml(1 hunks)src/unstract/sdk/__init__.py(1 hunks)
🔇 Additional comments (4)
src/unstract/sdk/__init__.py (1)
4-6: LGTM: stable API for version retrieval.
get_sdk_version()correctly returns the bumped version.pyproject.toml (3)
29-29: Major bump to llama-index 0.13.x — validate runtime compatibility across adapters.0.13 includes breaking/internal changes vs 0.12.x. Please run the full adapter matrix (OpenAI/Anthropic/Vertex/Bedrock, vector stores, embeddings) in CI to catch regressions.
I can generate a focused smoke-test script that imports each configured adapter and performs a 1-token call against a mock transport. Want that added to tests?
56-57: Bedrock and boto3 compatibility.
llama-index-llms-bedrock-converse==0.8.2plus extrasboto3~=1.34.131: confirm the Converse runtime features you rely on exist in this boto3 minor. If you need Bedrock Converse streaming/tool-use updates, you may require a newer boto3.I can add a small runtime check in an integration test to assert Converse model invocation works against a stubbed client.
Also applies to: 75-79
27-27: Dotenv minor bump — verify no behavior change with.envloading in your CLI entrypoints.
python-dotenv==1.0.1changed some deprecation paths versus 1.0.0. Sanity-checkload_dotenv()usage in CLI and tests.If needed, I’ll scan the repo to locate
dotenvusage and add a tiny smoke test.
gaya3-zipstack
left a comment
gaya3-zipstack
left a comment
There was a problem hiding this comment.
Changes looks good. Please test "test-connection" for all adapters with supported models!
chandrasekharan-zipstack
deleted the
UN-2725-add-support-for-open-ai-gpt-5-and-claude-sonnet-4-1-m-token-context
branch
3 participants
What
0.13.2GPT-5modelsWhy
How
Relevant Docs
Related Issues or PRs
Dependencies Versions / Env Variables
Notes on Testing
Screenshots
...
Checklist
I have read and understood the Contribution Guidelines.