Skip to content
Sign in
Talk to our team
Talk to our teamTry aixplain One
All posts
Release notes

2.10.5

2 min read

SDK

SDK build 0.2.34

SDK build 0.2.33

  • Improved session ID creation and management.

Assets

Models

  • Claude 4.5 Sonnet
  • DeepSeek V3.1
  • DeepSeek V3.1 Terminus
  • Gemini 2.5 Pro

Agent Engine

  • More focused inspector reviews. Inspectors now judge only the latest run of a step, without earlier critiques mixed in. A step that ran several times appears once, with its final output.
  • Steadier team runs. Team agents now follow the dependencies between the steps in their plan, and integration tools with several actions each run against the right connection. A run that hits the model's max tokens now says so.
  • Agents follow their description. An agent's description and instructions are now part of its system prompt, and agents in a team receive their own instructions.
  • Unique draft names. Creating a draft agent with the name of an agent you already have now returns an error.
  • Max tokens respected. The max tokens you set on an agent or team agent now reaches the model, including the agents in a team.

SDK updates

  • Custom inspector policies (private beta). An inspector's policy can now be a Python function as well as WARN, ABORT or ADAPTIVE. You can also use a tool, your own Python check, as an inspector, next to guardrail and text-generation models.
  • Index files directly. upsert() on an index now takes a file path, such as a text or PDF file, and parses the file into records for you.
  • Workflow tasks replace agent tasks. Use WorkflowTask with the workflow_tasks parameter and create_workflow_task(); the old tasks parameter still works but shows a deprecation warning. Creating a team from YAML no longer duplicates tasks.
  • One-step deployment. Deploying an agent or team agent now deploys its tools and sub-agents too, and skips any that are already deployed.
  • Clearer API key errors. AIXPLAIN_API_KEY and TEAM_API_KEY both work everywhere in the SDK, and errors no longer suggest that only TEAM_API_KEY is accepted.
  • Request IDs for tracing. Pass trace_request=True when you run an agent or team agent to log the run's request ID, which helps when you trace a run or ask for support.
  • Fixes. Listing agents and team agents works again when some of them are broken, AgentFactory no longer carries saved settings over from earlier calls, and an expected_output given as a Pydantic model is handled correctly.

Knowledge

  • Integrations in search. Marketplace search now finds integrations.