New Claude API Console Features Explained: Longer Outputs, Prompt Generation, and a Cost Dashboard

If you’ve recently been using the Claude API for customer support, content generation, or internal tools, you’ve probably noticed the Claude console has become significantly more usable. This update focuses on practical improvements: Claude Sonnet 3.5 supports longer outputs, the Workbench adds a prompt generator and an evaluation mode, and Claude API usage and costs can now be tracked by dimension in a dashboard. Below, I’ll break it down by feature so you can apply it directly.

Claude Sonnet 3.5 Long Output: Increased from 4096 to 8192

In the Claude API, Claude Sonnet 3.5’s maximum output token limit has doubled from 4096 to 8192, making long-form summaries, long code generation, and complex documentation smoother. To enable extended output, you need to add a beta request header to your request: anthropic-beta: "max-tokens-3-5-sonnet-2024-07-15". After enabling it, it’s recommended that you also control the length and structure of your prompt; otherwise, even with longer output, Claude’s response may become more “scattered.”

Claude Workbench Prompt Generator: Describe the task first, then let it write the prompt

The Claude console Workbench now includes a “prompt generator.” It works much like handing a colleague a requirements brief: you describe the task objective first (for example, “classify and handle incoming customer support requests”), and Claude then generates a more reusable prompt template. For teams, this helps reduce style inconsistency caused by “everyone writing their own prompts,” and it’s better suited for standardized workflows.

Evaluation Mode Is Live: Compare prompts side by side and score on a 5-point scale

The Workbench “evaluation mode” lets you run the same task side by side with two or more prompts, then rate Claude’s outputs on a 5-point scale. The pain point it solves isn’t “which one looks better,” but enabling you to reliably reproduce differences under the same input conditions. For prompt acceptance testing before launching with the Claude API, and for A/B iteration, it’s far more efficient than manually comparing results in a chat window.

Usage and Cost Dashboard: Track bills by USD, tokens, and API key

The developer console adds new “Usage” and “Costs” tabs, allowing you to track Claude API usage and billing by USD amount, token count, and API key. For teams with multiple environments (staging/production) or multiple projects sharing Claude API keys, this makes it faster to pinpoint abnormal consumption. The official documentation also adds a more complete entry point for release notes, making it easier to find what changed in future updates across Claude’s API, console, and apps.