Incorrect LLM model referenced in Insiders tab

BNAndras · September 29, 2026, 3:15pm

exercism/website/blob/489dd30906fb0aae15f698679ef5b11b6f8f8bce/config/locales/views/insiders/show.yml#L46


      
              next call.
          dark_theme:
            title: Dark theme
            description_html:
              We've put together a lovely Dark Theme for you to enjoy. Give
              your eyes a break while coding at night and enjoy Exercism in its
              darkest glory. You can toggle your theme in your <a
              href="%{user_preferences_link}">User Preferences</a>.
          chatgpt:
            title: ChatGPT
            description:
              Our ChatGPT integration will help you get unstuck directly in our
              online editor. You get 3x v4.0 requests and 30x v3.5 requests per
              calendar month. Let us know how you get on!
          name_tag_flair_and_badges:
            title: Name tag flair & Badges
            description_html:
              You'll now see a shiny little %{icon} next to your name around
              the site and a new tag on <a href="%{profile_link}">your public
              profile</a>. At some point soon, we'll be letting you customise which
              icon you'd like!

The LLM was recently replaced. I believe it’s Google Gemini Flash 2.5 now.

iHiD · September 29, 2026, 9:31pm

Thanks

3.5 actually!

BNAndras · September 29, 2026, 9:38pm

Gotcha.

Does this need to be updated too? It says 2.5, but I don’t know if that’s a throwaway reference.

github.com

exercism/website/blob/090bc5edfae3a66e669c34b5b207dffe9b3074e1/app/javascript/i18n/extract-jsx-copy/runLLM.ts#L8


      
          import { GoogleGenAI } from '@google/genai'
          
          export async function runLLM(prompt: string): Promise<string | undefined> {
            const apiKey = process.env.GOOGLE_GENAI_API_KEY
            const ai = new GoogleGenAI({ apiKey })
          
            const response = await ai.models.generateContent({
              model: 'gemini-2.5-flash',
              contents: prompt,
              config: {
                thinkingConfig: {
                  thinkingBudget: 0,
                },
              },
            })
          
            return response.text
          }