diff --git a/config/ai.ts b/config/ai.ts
index 5b653aa..5f9a5cf 100644
--- a/config/ai.ts
+++ b/config/ai.ts
@@ -2,7 +2,7 @@ import type { ModelInfo } from "../src/commands/ai"
 
 export const defaultFlashModel = "gemma3:4b"
 export const defaultThinkingModel = "qwen3:4b"
-export const unloadModelAfterB = 0.1 // how many billion params until model is auto-unloaded
+export const unloadModelAfterB = 12 // how many billion params until model is auto-unloaded
 
 export const models: ModelInfo[] = [
   {
@@ -35,13 +35,31 @@ export const models: ModelInfo[] = [
     descriptionPt: 'Qwen3 é uma série de modelos multilingues.',
     models: [
       { name: 'qwen3:0.6b', label: 'Qwen3 0.6B', parameterSize: '0.6B' },
+      { name: 'qwen3:1.7b', label: 'Qwen3 1.7B', parameterSize: '1.7B' },
       { name: 'qwen3:4b', label: 'Qwen3 4B', parameterSize: '4B' },
       { name: 'qwen3:8b', label: 'Qwen3 8B', parameterSize: '8B' },
       { name: 'qwen3:14b', label: 'Qwen3 14B', parameterSize: '14B' },
       { name: 'qwen3:30b', label: 'Qwen3 30B', parameterSize: '30B' },
+      { name: 'qwen3:32b', label: 'Qwen3 32B', parameterSize: '32B' },
       { name: 'qwen3:235b-a22b', label: 'Qwen3 235B A22B', parameterSize: '235B' },
     ]
   },
+  {
+    name: 'qwen3-abliterated',
+    label: 'Qwen3   [ Uncensored ]',
+    descriptionEn: 'Qwen3-abliterated is a multilingual reasoning model series.',
+    descriptionPt: 'Qwen3-abliterated é uma série de modelos multilingues.',
+    models: [
+      { name: 'huihui_ai/qwen3-abliterated:0.6b', label: 'Qwen3 Uncensored 0.6B', parameterSize: '0.6B' },
+      { name: 'huihui_ai/qwen3-abliterated:1.7b', label: 'Qwen3 Uncensored 1.7B', parameterSize: '1.7B' },
+      { name: 'huihui_ai/qwen3-abliterated:4b', label: 'Qwen3 Uncensored 4B', parameterSize: '4B' },
+      { name: 'huihui_ai/qwen3-abliterated:8b', label: 'Qwen3 Uncensored 8B', parameterSize: '8B' },
+      { name: 'huihui_ai/qwen3-abliterated:14b', label: 'Qwen3 Uncensored 14B', parameterSize: '14B' },
+      { name: 'huihui_ai/qwen3-abliterated:30b', label: 'Qwen3 Uncensored 30B', parameterSize: '30B' },
+      { name: 'huihui_ai/qwen3-abliterated:32b', label: 'Qwen3 Uncensored 32B', parameterSize: '32B' },
+      { name: 'huihui_ai/qwen3-abliterated:235b', label: 'Qwen3 Uncensored 235B', parameterSize: '235B' },
+    ]
+  },
   {
     name: 'qwq',
     label: 'QwQ',
@@ -49,6 +67,7 @@ export const models: ModelInfo[] = [
     descriptionPt: 'QwQ é o modelo de raciocínio da série Qwen.',
     models: [
       { name: 'qwq:32b', label: 'QwQ 32B', parameterSize: '32B' },
+      { name: 'huihui_ai/qwq-abliterated:32b', label: 'QwQ Uncensored 32B', parameterSize: '32B' },
     ]
   },
   {
@@ -60,6 +79,32 @@ export const models: ModelInfo[] = [
       { name: 'llama4:scout', label: 'Llama4 109B A17B', parameterSize: '109B' },
     ]
   },
+  {
+    name: 'deepseek',
+    label: 'DeepSeek   [ & Uncensored ]',
+    descriptionEn: 'DeepSeek is a research model for reasoning tasks.',
+    descriptionPt: 'DeepSeek é um modelo de pesquisa para tarefas de raciocínio.',
+    models: [
+      { name: 'deepseek-r1:1.5b', label: 'DeepSeek 1.5B', parameterSize: '1.5B' },
+      { name: 'deepseek-r1:7b', label: 'DeepSeek 7B', parameterSize: '7B' },
+      { name: 'deepseek-r1:8b', label: 'DeepSeek 8B', parameterSize: '8B' },
+      { name: 'deepseek-r1:14b', label: 'DeepSeek 14B', parameterSize: '14B' },
+      { name: 'huihui_ai/deepseek-r1-abliterated:1.5b', label: 'DeepSeek Uncensored 1.5B', parameterSize: '1.5B' },
+      { name: 'huihui_ai/deepseek-r1-abliterated:7b', label: 'DeepSeek Uncensored 7B', parameterSize: '7B' },
+      { name: 'huihui_ai/deepseek-r1-abliterated:8b', label: 'DeepSeek Uncensored 8B', parameterSize: '8B' },
+      { name: 'huihui_ai/deepseek-r1-abliterated:14b', label: 'DeepSeek Uncensored 14B', parameterSize: '14B' },
+    ]
+  },
+  {
+    name: 'hermes3',
+    label: 'Hermes3',
+    descriptionEn: 'Hermes 3 is the latest version of the flagship Hermes series of LLMs by Nous Research.',
+    descriptionPt: 'Hermes 3 é a versão mais recente da série Hermes de LLMs da Nous Research.',
+    models: [
+      { name: 'hermes3:3b', label: 'Hermes3 3B', parameterSize: '3B' },
+      { name: 'hermes3:8b', label: 'Hermes3 8B', parameterSize: '8B' },
+    ]
+  },
   {
     name: 'mistral',
     label: 'Mistral',
@@ -70,15 +115,15 @@ export const models: ModelInfo[] = [
     ]
   },
   {
-    name: 'deepseek',
-    label: 'DeepSeek   [ & Uncensored ]',
-    descriptionEn: 'DeepSeek is a research model for reasoning tasks.',
-    descriptionPt: 'DeepSeek é um modelo de pesquisa para tarefas de raciocínio.',
+    name: 'phi4   [ & Uncensored ]',
+    label: 'Phi4',
+    descriptionEn: 'Phi-4 is a 14B parameter, state-of-the-art open model from Microsoft. ',
+    descriptionPt: 'Phi-4 é um modelo de 14B de última geração, aberto pela Microsoft.',
     models: [
-      { name: 'deepseek-r1:1.5b', label: 'DeepSeek 1.5B', parameterSize: '1.5B' },
-      { name: 'deepseek-r1:7b', label: 'DeepSeek 7B', parameterSize: '7B' },
-      { name: 'huihui_ai/deepseek-r1-abliterated:1.5b', label: 'DeepSeek Uncensored 1.5B', parameterSize: '1.5B' },
-      { name: 'huihui_ai/deepseek-r1-abliterated:7b', label: 'DeepSeek Uncensored 7B', parameterSize: '7B' },
+      { name: 'hf.co/unsloth/Phi-4-mini-reasoning-GGUF', label: 'Phi4 Mini Reasoning', parameterSize: '4B' },
+      { name: 'phi4:14b', label: 'Phi4 14B', parameterSize: '14B' },
+      { name: 'hf.co/unsloth/Phi-4-reasoning-plus-GGUF', label: 'Phi4 Reasoning Plus', parameterSize: '14B' },
+      { name: 'huihui_ai/phi4-abliterated:14b', label: 'Phi4 Uncensored 14B', parameterSize: '14B' },
     ]
   },
   {
@@ -89,5 +134,34 @@ export const models: ModelInfo[] = [
     models: [
       { name: 'phi3:3.8b', label: 'Phi3 3.8B', parameterSize: '3.8B' },
     ]
-  }
+  },
+  {
+    name: 'llama3',
+    label: 'Llama4',
+    descriptionEn: 'Llama 3, a lightweight model from Meta.',
+    descriptionPt: 'Llama 3, um modelo leve da Meta.',
+    models: [
+      { name: 'llama3:8b', label: 'Llama3 8B', parameterSize: '8B' },
+    ]
+  },
+  {
+    name: 'llama3.1   [ Uncensored ]',
+    label: 'Llama3.1',
+    descriptionEn: 'Ablitered v3 llama-3.1 8b with uncensored prompt ',
+    descriptionPt: 'Llama3.1 é um modelo aberto, leve e para dispositivos locais, com prompt não censurado.',
+    models: [
+      { name: 'mannix/llama3.1-8b-abliterated:latest', label: 'Llama3.1 8B', parameterSize: '8B' },
+    ]
+  },
+  {
+    name: 'llama3.2   [ & Uncensored ]',
+    label: 'Llama3.2',
+    descriptionEn: 'Llama3.2 is a family of open, lightweight models for general tasks.',
+    descriptionPt: 'Llama3.2 é uma família de modelos abertos, leves e para dispositivos locais, para tarefas gerais.',
+    models: [
+      { name: 'llama3.2:1b', label: 'Llama3.2 1B', parameterSize: '1B' },
+      { name: 'llama3.2:3b', label: 'Llama3.2 3B', parameterSize: '3B' },
+      { name: 'socialnetwooky/llama3.2-abliterated:3b_q8_0', label: 'Llama3.2 Uncensored 3B', parameterSize: '3B' },
+    ]
+  },
 ];
\ No newline at end of file
diff --git a/config/settings.ts b/config/settings.ts
new file mode 100644
index 0000000..1fe94b3
--- /dev/null
+++ b/config/settings.ts
@@ -0,0 +1,2 @@
+export const seriesPageSize = 4;
+export const modelPageSize = 4;
\ No newline at end of file
diff --git a/nodemon.json b/nodemon.json
index 918bcb8..4e4ec20 100644
--- a/nodemon.json
+++ b/nodemon.json
@@ -1,6 +1,6 @@
-{   
+{
   "ignore": ["src/props/*.json", "src/props/*.txt"],
-  "watch": ["src"],
+  "watch": ["src", "config"],
   "ext": "ts,js",
   "exec": "bun src/bot.ts"
 }
\ No newline at end of file
diff --git a/src/commands/ai.ts b/src/commands/ai.ts
index ebcd613..a032850 100644
--- a/src/commands/ai.ts
+++ b/src/commands/ai.ts
@@ -135,6 +135,26 @@ function sanitizeMarkdownForTelegram(text: string): string {
   return sanitizedText;
 }
 
+function processThinkingTags(text: string): string {
+  let processedText = text;
+
+  const firstThinkIndex = processedText.indexOf('<think>');
+  if (firstThinkIndex === -1) {
+      return processedText.replace(/<\/think>/g, '___THINK_END___');
+  }
+
+  processedText = processedText.substring(0, firstThinkIndex) + '___THINK_START___' + processedText.substring(firstThinkIndex + '<think>'.length);
+  const lastThinkEndIndex = processedText.lastIndexOf('</think>');
+  if (lastThinkEndIndex !== -1) {
+      processedText = processedText.substring(0, lastThinkEndIndex) + '___THEND___' + processedText.substring(lastThinkEndIndex + '</think>'.length);
+  }
+  processedText = processedText.replace(/<think>/g, '');
+  processedText = processedText.replace(/<\/think>/g, '');
+  processedText = processedText.replace('___THEND___', '___THINK_END___');
+
+  return processedText;
+}
+
 export async function preChecks() {
   const envs = [
     "ollamaApi",
@@ -221,9 +241,10 @@ async function getResponse(prompt: string, ctx: TextContext, replyGenerating: Me
       error: Strings.unexpectedErr.replace("{error}", Strings.ai.noChatFound),
     };
   }
+  const cleanedModelName = model.replace('hf.co/', '');
   let status = Strings.ai.statusWaitingRender;
   let modelHeader = Strings.ai.modelHeader
-    .replace("{model}", model)
+    .replace("{model}", cleanedModelName)
     .replace("{temperature}", aiTemperature)
     .replace("{status}", status) + "\n\n";
 
@@ -277,23 +298,8 @@ async function getResponse(prompt: string, ctx: TextContext, replyGenerating: Me
           console.error("[✨ AI | !] Error parsing chunk");
           continue;
         }
-        if (model === thinking_model && !showThinking) {
-          if (ln.response) {
-            finalResponseText += ln.response;
-            if (finalResponseText.includes('<think>') && !thinkingMessageSent) {
-              await rateLimiter.editMessageWithRetry(
-                ctx,
-                ctx.chat.id,
-                replyGenerating.message_id,
-                modelHeader + Strings.ai.thinking,
-                { parse_mode: 'Markdown' }
-              );
-              thinkingMessageSent = true;
-            }
-          }
-          continue;
-        }
-        if (model === thinking_model && ln.response) {
+
+        if (ln.response) {
           if (ln.response.includes('<think>')) {
             const thinkMatch = ln.response.match(/<think>([\s\S]*?)<\/think>/);
             if (thinkMatch && thinkMatch[1].trim().length > 0) {
@@ -304,68 +310,63 @@ async function getResponse(prompt: string, ctx: TextContext, replyGenerating: Me
           } else if (ln.response.includes('</think>')) {
             logger.logThinking(ctx.chat.id, replyGenerating.message_id, false);
           }
-        }
-        if (ln.response) {
-          if (model === thinking_model) {
-            let patchedThoughts = ln.response;
-            const thinkTagRx = /<think>([\s\S]*?)<\/think>/g;
-            patchedThoughts = patchedThoughts.replace(thinkTagRx, (p1) => p1.trim().length > 0 ? '___THINK_START___' + p1.trim() + '___THINK_END___' : '');
-            patchedThoughts = patchedThoughts.replace(/<think>/g, '___THINK_START___');
-            patchedThoughts = patchedThoughts.replace(/<\/think>/g, '___THINK_END___');
-            thoughts += patchedThoughts;
-            fullResponse += patchedThoughts;
-          } else {
-            fullResponse += ln.response;
-          }
-          if (firstChunk) {
-            status = Strings.ai.statusWaitingRender;
-            modelHeader = Strings.ai.modelHeader
-              .replace("{model}", model)
-              .replace("{temperature}", aiTemperature)
-              .replace("{status}", status) + "\n\n";
-            await rateLimiter.editMessageWithRetry(
-              ctx,
-              ctx.chat.id,
-              replyGenerating.message_id,
-              modelHeader + formatThinkingMessage(fullResponse),
-              { parse_mode: 'Markdown' }
-            );
-            lastUpdateCharCount = fullResponse.length;
-            sentHeader = true;
-            firstChunk = false;
-            continue;
-          }
-          const updateEveryChars = Number(process.env.updateEveryChars) || 100;
-          if (fullResponse.length - lastUpdateCharCount >= updateEveryChars || !sentHeader) {
-            await rateLimiter.editMessageWithRetry(
-              ctx,
-              ctx.chat.id,
-              replyGenerating.message_id,
-              modelHeader + formatThinkingMessage(fullResponse),
-              { parse_mode: 'Markdown' }
-            );
-            lastUpdateCharCount = fullResponse.length;
-            sentHeader = true;
+          fullResponse += ln.response;
+          if (showThinking) {
+            let displayResponse = processThinkingTags(fullResponse);
+
+            if (firstChunk) {
+              status = Strings.ai.statusWaitingRender;
+              modelHeader = Strings.ai.modelHeader
+                .replace("{model}", cleanedModelName)
+                .replace("{temperature}", aiTemperature)
+                .replace("{status}", status) + "\n\n";
+              await rateLimiter.editMessageWithRetry(
+                ctx,
+                ctx.chat.id,
+                replyGenerating.message_id,
+                modelHeader + formatThinkingMessage(displayResponse),
+                { parse_mode: 'Markdown' }
+              );
+              lastUpdateCharCount = displayResponse.length;
+              sentHeader = true;
+              firstChunk = false;
+              continue;
+            }
+            const updateEveryChars = Number(process.env.updateEveryChars) || 100;
+            if (displayResponse.length - lastUpdateCharCount >= updateEveryChars || !sentHeader) {
+              await rateLimiter.editMessageWithRetry(
+                ctx,
+                ctx.chat.id,
+                replyGenerating.message_id,
+                modelHeader + formatThinkingMessage(displayResponse),
+                { parse_mode: 'Markdown' }
+              );
+              lastUpdateCharCount = displayResponse.length;
+              sentHeader = true;
+            }
           }
         }
       }
     }
-    if (model === thinking_model && !showThinking) {
-      const cleanedResponse = finalResponseText.replace(/<think>[\s\S]*?<\/think>/g, '').trim();
-      return { success: true, response: cleanedResponse };
-    }
+
     status = Strings.ai.statusRendering;
     modelHeader = Strings.ai.modelHeader
-      .replace("{model}", model)
+      .replace("{model}", cleanedModelName)
       .replace("{temperature}", aiTemperature)
       .replace("{status}", status) + "\n\n";
-    await rateLimiter.editMessageWithRetry(
-      ctx,
-      ctx.chat.id,
-      replyGenerating.message_id,
-      modelHeader + formatThinkingMessage(fullResponse),
-      { parse_mode: 'Markdown' }
-    );
+
+    if (showThinking) {
+        let displayResponse = processThinkingTags(fullResponse);
+
+        await rateLimiter.editMessageWithRetry(
+            ctx,
+            ctx.chat.id,
+            replyGenerating.message_id,
+            modelHeader + formatThinkingMessage(displayResponse),
+            { parse_mode: 'Markdown' }
+        );
+    }
+
     const responseCharCount = fullResponse.length;
     await db.update(schema.usersTable)
       .set({
@@ -373,9 +374,12 @@ async function getResponse(prompt: string, ctx: TextContext, replyGenerating: Me
         aiRequests: sql`${schema.usersTable.aiRequests} + 1`
       })
       .where(eq(schema.usersTable.telegramId, userId));
+
+    const patchedResponse = processThinkingTags(fullResponse);
+
     return {
       success: true,
-      response: fullResponse,
+      response: patchedResponse,
     };
   } catch (error: unknown) {
     const errorMsg = extractAxiosErrorMessage(error);
@@ -429,16 +433,20 @@ async function handleAiReply(ctx: TextContext, model: string, prompt: string, re
   if (!aiResponse) return;
   if (!ctx.chat) return;
   if (aiResponse.success && aiResponse.response) {
+    const cleanedModelName = model.replace('hf.co/', '');
     const status = Strings.ai.statusComplete;
     const modelHeader = Strings.ai.modelHeader
-      .replace("{model}", model)
+      .replace("{model}", cleanedModelName)
       .replace("{temperature}", aiTemperature)
       .replace("{status}", status) + "\n\n";
     const urlWarning = containsUrls(originalMessage) ? Strings.ai.urlWarning : '';
     let finalResponse = aiResponse.response;
-    if (model === thinking_model) {
-      finalResponse = finalResponse.replace(/___THINK_START___/g, `${Strings.ai.thinking}`)
-                                 .replace(/___THINK_END___/g, `${Strings.ai.finishedThinking}`);
+    if (showThinking) {
+        finalResponse = finalResponse.replace(/___THINK_START___/g, `${Strings.ai.thinking}`)
+                                     .replace(/___THINK_END___/g, `${Strings.ai.finishedThinking}`);
+    } else {
+        finalResponse = finalResponse.replace(/___THINK_START___[\s\S]*?___THINK_END___/g, '').trim();
+        finalResponse = finalResponse.replace(/___THINK_START___[\s\S]*/g, '').trim();
     }
 
     await rateLimiter.editMessageWithRetry(
diff --git a/src/commands/main.ts b/src/commands/main.ts
index 9478a13..f86ddea 100644
--- a/src/commands/main.ts
+++ b/src/commands/main.ts
@@ -10,6 +10,7 @@ import type { NodePgDatabase } from 'drizzle-orm/node-postgres';
 import { getModelLabelByName } from './ai';
 import { models } from '../../config/ai';
 import { langs } from '../locales/config';
+import { modelPageSize, seriesPageSize } from '../../config/settings';
 
 type UserRow = typeof schema.usersTable.$inferSelect;
 
@@ -221,12 +222,12 @@ export default (bot: Telegraf<Context>, db: NodePgDatabase<typeof schema>) => {
       const originalIndex = start + idx;
       const isSelected = series.models.some(m => m.name === user.customAiModel);
       const label = isSelected ? `✅ ${series.label}` : series.label;
-      return { text: label, callback_data: `selectseries_${originalIndex}_${user.telegramId}` };
+      return { text: label, callback_data: `selectseries_${originalIndex}_0_${user.telegramId}` };
     });
 
     const navigationButtons: any[] = [];
     if (page > 0) {
-      navigationButtons.push({ text: Strings.varStrings.varBack, callback_data: `settings_aiModel_${page - 1}_${user.telegramId}` });
+      navigationButtons.push({ text: Strings.varStrings.varLess, callback_data: `settings_aiModel_${page - 1}_${user.telegramId}` });
     }
     if (end < models.length) {
       navigationButtons.push({ text: Strings.varStrings.varMore, callback_data: `settings_aiModel_${page + 1}_${user.telegramId}` });
@@ -257,7 +258,7 @@ export default (bot: Telegraf<Context>, db: NodePgDatabase<typeof schema>) => {
     }
   });
 
-  bot.action(/^selectseries_\d+_\d+$/, async (ctx) => {
+  bot.action(/^selectseries_\d+_\d+_\d+$/, async (ctx) => {
     const data = (ctx.callbackQuery as any).data;
     const userId = extractUserIdFromCallback(data);
     const allowed = !!userId && String(ctx.from.id) === userId;
@@ -269,26 +270,46 @@ export default (bot: Telegraf<Context>, db: NodePgDatabase<typeof schema>) => {
     await ctx.answerCbQuery();
     const { user, Strings } = await getUserAndStrings(ctx, db);
     if (!user) return;
-    const match = data.match(/^selectseries_(\d+)_\d+$/);
+    const match = data.match(/^selectseries_(\d+)_(\d+)_(\d+)$/);
     if (!match) return;
     const seriesIdx = parseInt(match[1], 10);
+    const modelPage = parseInt(match[2], 10);
     const series = models[seriesIdx];
     if (!series) return;
-    const pageSize = 4;
-    const page = Math.floor(seriesIdx / pageSize);
+
+    const seriesPage = Math.floor(seriesIdx / seriesPageSize);
+
+    const start = modelPage * modelPageSize;
+    const end = start + modelPageSize;
+    const paginatedSeriesModels = series.models.slice(start, end);
+
+    const modelButtons = paginatedSeriesModels.map((m, idx) => {
+      const originalModelIndex = start + idx;
+      const isSelected = m.name === user.customAiModel;
+      const label = isSelected ? `✅ ${m.label}` : m.label;
+      return [{ text: `${label} (${m.parameterSize})`, callback_data: `setmodel_${seriesIdx}_${originalModelIndex}_${user.telegramId}` }];
+    });
+
+    const navigationButtons: any[] = [];
+    if (modelPage > 0) {
+      navigationButtons.push({ text: Strings.varStrings.varLess, callback_data: `selectseries_${seriesIdx}_${modelPage - 1}_${user.telegramId}` });
+    }
+    if (end < series.models.length) {
+      navigationButtons.push({ text: Strings.varStrings.varMore, callback_data: `selectseries_${seriesIdx}_${modelPage + 1}_${user.telegramId}` });
+    }
+
+    const keyboard: any[][] = [...modelButtons];
+    if (navigationButtons.length > 0) {
+      keyboard.push(navigationButtons);
+    }
+    keyboard.push([{ text: `${Strings.varStrings.varBack}`, callback_data: `settings_aiModel_${seriesPage}_${user.telegramId}` }]);
     const desc = user.languageCode === 'pt' ? series.descriptionPt : series.descriptionEn;
     try {
       await ctx.editMessageText(
         `${Strings.settings.ai.seriesDescription.replace('{seriesDescription}', desc)}\n\n${Strings.settings.ai.selectParameterSize.replace('{seriesLabel}', series.label).replace('   [ & Uncensored ]', '')}\n\n${Strings.settings.ai.parameterSizeExplanation}`,
         {
           reply_markup: {
-            inline_keyboard: series.models.map((m, idx) => {
-              const isSelected = m.name === user.customAiModel;
-              const label = isSelected ? `✅ ${m.label}` : m.label;
-              return [{ text: `${label} (${m.parameterSize})`, callback_data: `setmodel_${seriesIdx}_${idx}_${user.telegramId}` }];
-            }).concat([[
-              { text: `${Strings.varStrings.varBack}`, callback_data: `settings_aiModel_${page}_${user.telegramId}` }
-            ]])
+            inline_keyboard: keyboard
           }
         }
       );
diff --git a/src/locales/english.json b/src/locales/english.json
index 0f6f875..dcb178d 100644
--- a/src/locales/english.json
+++ b/src/locales/english.json
@@ -80,7 +80,7 @@
     "inQueue": "ℹ️ You are {position} in the queue.",
     "startingProcessing": "✨ Starting to process your request...",
     "systemPrompt": "You are a friendly assistant called {botName}.\nCurrent Date/Time (UTC): {date}\n\n---\n\nUser message:\n{message}",
-    "statusWaitingRender": "⏳ Waiting to Render...",
+    "statusWaitingRender": "⏳ Streaming...",
     "statusRendering": "🖼️ Rendering...",
     "statusComplete": "✅ Complete!",
     "modelHeader": "🤖 *{model}*      🌡️  *{temperature}*      {status}",
@@ -123,7 +123,7 @@
       "aiEnabledSetTo": "AI Enabled set to {aiEnabled}",
       "aiModelSetTo": "AI Model set to {aiModel}",
       "aiTemperatureSetTo": "AI Temperature set to {aiTemperature}",
-      "selectSeries": "*Please select a model series.*",
+      "selectSeries": "*Please select a model series.*\n\nThis will be set as the default model for the /ai command.",
       "seriesDescription": "{seriesDescription}",
       "selectParameterSize": "*Please select a parameter size for {seriesLabel}*.",
       "parameterSizeExplanation": "Parameter size (e.g. 2B, 4B) refers to the number of parameters in the model. Larger models may be more capable but require more resources.",
diff --git a/src/locales/portuguese.json b/src/locales/portuguese.json
index 6571b12..8697c2b 100644
--- a/src/locales/portuguese.json
+++ b/src/locales/portuguese.json
@@ -85,7 +85,7 @@
         "selectParameterSize": "Por favor, selecione um tamanho de parâmetro para {seriesLabel}.",
         "parameterSizeExplanation": "O tamanho do parâmetro (ex: 2B, 4B) refere-se ao número de parâmetros do modelo. Modelos maiores podem ser mais capazes, mas exigem mais recursos.",
         "systemPrompt": "Você é um assistente de Telegram chamado {botName}.\nData/Hora atual (UTC): {date}\n\n---\n\nMensagem do usuário:\n{message}",
-        "statusWaitingRender": "⏳ Aguardando renderização...",
+        "statusWaitingRender": "⏳ Transmitindo...",
         "statusRendering": "🖼️ Renderizando...",
         "statusComplete": "✅ Completo!",
         "modelHeader": "🤖 *{model}*      🌡️  *{temperature}*      {status}",
@@ -126,7 +126,7 @@
             "aiEnabledSetTo": "Inteligência Artificial definido para {aiEnabled}",
             "aiModelSetTo": "Modelo personalizado definido para {aiModel}",
             "aiTemperatureSetTo": "Temperatura definida para {aiTemperature}",
-            "selectSeries": "*Por favor, selecione uma série de modelos.*",
+            "selectSeries": "*Por favor, selecione uma série de modelos.*\n\nIsso será definido como o modelo padrão para o comando /ai.",
             "seriesDescription": "{seriesDescription}",
             "selectParameterSize": "Por favor, selecione um tamanho de parâmetro para {seriesLabel}.",
             "parameterSizeExplanation": "O tamanho do parâmetro (ex: 2B, 4B) refere-se ao número de parâmetros do modelo. Modelos maiores podem ser mais capazes, mas exigem mais recursos.",