Maxim AI release notes
Maxim AI release notes www.getmaxim.ai

📥 Import cURL directly as Workflow

 

New

  

You can directly import cURL into Maxim by pasting it in the address bar.

curl-import.gif

📋 Trigger test runs using NodeJS SDK

 

New

  

With the Maxim SDK, you can now trigger test runs to evaluate our workflows, enabling us to:

  • Use local CSV files or other data sources as test datasets
  • Test workflows running on our local machine without bringing them to Maxim.
  • Track test run status through the Maxim dashboard, just like regular runs

Learn more here.

code-test-run.gif

💰 Evaluation cost column in the test runs

 

New

  
  • We have added per-entry evaluation costs in test run reports when using LLM judge evaluators.
  • The cost column supports filtering and sorting operations.

evaluation cost.gif

🧊 We now support INFERENCE_PROFILE inference type on Bedrock

 

New

  
  • We’ve added support for the INFERENCE_PROFILE inference type on Bedrock, which enables region-specific model deployments on AWS Bedrock 🎉.
  • To add these models, you must provide the Inference profile ID and the corresponding ARN.

inference-models.gif

📤 Export CSV of logs along with evaluation results using filters

 

New

  

Logs export now includes evaluation data in CSV format. Users can download both logs and their linked evaluations in a single export.

How to Use

  • Go to the Logs page
  • Click "Export CSV" button
  • Download includes:
    • Log entries
    • Related evaluation metrics
    • Scores and feedback

export-logs.gif

0️⃣ Normalize evaluator scores

 

New

  
  • For simplicity and accuracy of custom LLM as a judge evaluator, we configure scores on a scale of 1-5 if it's a numeric score.
  • But for reporting purposes, you may want to normalize the score between 0-1.
  • With this release, we have added an extra setting on scale-based LLM as a judge evaluator to enable normalizing the scores across the test runs and online evaluation.

normalization.gif

🪵 Logs and online evaluation views are merged

 

New

  
  • Based on the feedback and for our upcoming agentic changes, we have merged the logs table and online evaluation tables.
  • You can filter logs with evaluation using the omnibar.
  • And log exports will now include evaluations if present.

🪪 Customize PII entities based on your needs

 

New

  

This change adds a skip list to Maxim's PII evaluator configuration. You can customize which entities you want us to flag as PII.

How to customize

  1. Click on "Settings" on the PII evaluator page.
  2. Select entities you want to classify as PII.
  3. Save.

pii-gf.gif

🧊 Amazon Nova models are now available on Maxim

 

New

  

We've added support for Amazon's new Nova family and Titan models.

Amazon Nova family models

  • Amazon Nova Lite (1.0)
  • Amazon Nova Micro (1.0)
  • Amazon Nova Pro (1.0)

Amazon Titan family models

  • Amazon Titan Text G1 - Lite
  • Amazon Titan Text G1 - Express
  • Amazon Titan Text G1 - Premier

Arc.gif

📣 Alerting and Notifications

 

New

  

We now offer enhanced Alerting and Notifications, enabling seamless integrations with tools like Slack or PagerDuty to keep your team informed and proactive.

Alerts and notifs.gif

Here’s what you can do:

  • Integrate with your tools: Set up integrations for Slack or PagerDuty to receive alerts.
  • Test run notifications: Receive messages on your chosen Slack channel whenever a test run reaches the selected status (e.g., starts running, is completed, fails, queued, stopped, etc.).

Stay informed and responsive with Alerting and Notifications, ensuring your team is always aware.