Full BYOK Support & AI Credit Transparency

Version: 1.25.0, Date: May 14, 2026

BYOK support now works correctly across every AI-powered feature in EnginifyAI, and every AI action gives you clear, accurate feedback showing exactly what was consumed and how. This release closes a set of gaps where BYOK users were still being charged platform credits, adds usage transparency to features that previously ran silently, and fixes several newer OpenAI models that had stopped working entirely.

Highlights

• BYOK is now honored everywhere: When you have a Bring Your Own Key configured for the active provider, your platform credits are no longer touched.
• Newer OpenAI models work again: GPT-5 variants, o1, o1-pro, o4, Codex, and computer-use models now generate correctly.
• Per-request usage feedback across every AI feature: See either the exact tokens processed through your key, or the credits used and remaining, after every generation.
• Trial users get Explorer-level BYOK: Trials now include the same BYOK access as paid Explorer subscribers.

URL-to-Prompt Kit

• New: A model selector on the configure screen lets you choose which AI model processes your URL before you run it. Previously the model was fixed and could not be changed from this screen, so you had no way to match the model to the task or to your budget from inside the Kit.

• Improved: The Kit now uses the standard notification system for both errors and successful generations. It previously relied on browser-native alert dialogs for error messages and gave no feedback at all when a generation succeeded. Feedback now matches the rest of the platform.

• Fixed: BYOK users running the URL-to-Prompt Kit are no longer charged platform credits when a BYOK key is configured for the active provider. Requests now route through your key as intended.

Workflows

• Improved: Workflow summary generation now shows the same per-request usage notification as every other AI action, including step execution and step generation. Previously it could run without any confirmation of what it used.

• Fixed: Workflow summary generation now checks your quota before running and respects your BYOK configuration. It previously had no pre-flight usage check and would attempt to run regardless of your balance or key configuration.
• Fixed: BYOK users running Workflow steps, step generation, or summaries are no longer charged platform credits when a BYOK key is configured for the active provider.

Designer

• Improved: The Designer now uses the same notification style as the rest of the app. It had been running on a legacy style while every other page had moved to the current standard, so AI action feedback there looked and behaved differently than everywhere else.

• Fixed: BYOK users generating in the Designer are no longer charged platform credits when a BYOK key is configured for the active provider.

Builders

• Fixed: BYOK users running generations in the Prompt Builders and Intent Builder are no longer charged platform credits when a BYOK key is configured for the active provider.

Account & Billing

• Improved: Usage limits for libraries, prompts, prompt versions, and workflows now track your cumulative lifetime total rather than a resetting monthly window. Your usage numbers now match how the limits are defined on each plan, so what you see reflects the true total you have created since joining.

• Fixed: Users on an active trial now receive Explorer-level BYOK access for the duration of the trial, matching paid Explorer subscribers. Previously trials did not include BYOK access, which meant the trial did not accurately reflect what a paid Explorer plan offers.

Across the App

• Improved: Every AI action now shows a contextual usage notification the moment it completes. If you use your own API key, you see the tokens processed through your key, for example “🔑 4,200 tokens processed with your API key.” If you are on platform credits, you see the credits used and your remaining balance, for example “📊 12 AI credits used • 488 / 500 remaining.” These appear consistently across the Builders, Analyzers, Workflows, Designer, Intent Builder, and URL-to-Prompt Kit, where previously many surfaces showed only a generic success message or no credit information at all.
• Improved: Credit amounts now display as whole numbers throughout the platform. You will see “12 AI credits used” instead of “12.00 AI credits used,” which makes usage easier to read and track at a glance.
• Improved: AI requests now put less load on our systems on every generation. Each AI action previously required three separate database calls to calculate and record credit usage; those have been consolidated into a single call by threading the calculated value through the generation pipeline. You get the same accurate usage tracking with less overhead on every request.

• Fixed: Newer OpenAI models that route through the Responses API, including GPT-5 variants, o1, o1-pro, o4, Codex, and computer-use models, now generate correctly. A parameter wiring issue in how EnginifyAI communicated with the OpenAI API had been causing these models to return errors and produce no output.