diff --git a/bun.lock b/bun.lock
index 20e895e7..9ece8757 100644
--- a/bun.lock
+++ b/bun.lock
@@ -1,5 +1,6 @@
 {
   "lockfileVersion": 1,
+  "configVersion": 0,
   "workspaces": {
     "": {
       "name": "copilot-api",
diff --git a/src/lib/model-consumption.json b/src/lib/model-consumption.json
new file mode 100644
index 00000000..98a406d1
--- /dev/null
+++ b/src/lib/model-consumption.json
@@ -0,0 +1,21 @@
+{
+"models": [
+{"name": "GPT-4.1", "consumption": "0x"},
+{"name": "GPT-4o", "consumption": "0x"},
+{"name": "GPT-5 mini", "consumption": "0x"},
+{"name": "Grok Code Fast 1", "consumption": "0x"},
+{"name": "Claude Haiku 4.5", "consumption": "0.33x"},
+{"name": "Claude Sonnet 4", "consumption": "1x"},
+{"name": "Claude Sonnet 4.5", "consumption": "1x"},
+{"name": "Gemini 2.5 Pro", "consumption": "1x"},
+{"name": "GPT-5", "consumption": "1x"},
+{"name": "GPT-5-Codex (Preview)", "consumption": "1x"},
+{"name": "GPT-5.1", "consumption": "1x"},
+{"name": "GPT-5.1-Codex", "consumption": "1x"},
+{"name": "GPT-5.1-Codex-Mini", "consumption": "0.33x"},
+{"name": "Gemini 3 Pro (Preview)", "consumption": "1x"},
+{"name": "Raptor mini (Preview)", "consumption": "0x"},
+{"name": "GPT-5.1-Codex-Max (Preview)", "consumption": "1x"},
+{"name": "Claude Opus 4.5 (Preview)", "consumption": "1x"}
+]
+}
\ No newline at end of file
diff --git a/src/lib/model-matcher.ts b/src/lib/model-matcher.ts
new file mode 100644
index 00000000..ebf8894f
--- /dev/null
+++ b/src/lib/model-matcher.ts
@@ -0,0 +1,277 @@
+import consola from "consola"
+
+import modelConsumptionData from "./model-consumption.json"
+import { state } from "./state"
+
+/**
+ * 获取模型消耗值
+ * 从配置文件中查询指定模型的消耗系数
+ * 
+ * @param modelName - 模型名称
+ * @returns 消耗系数（如 1.0, 2.0 等），未找到或解析失败返回 999
+ * 
+ * @example
+ * getModelConsumption("claude-3.5-sonnet") // 返回 1.0
+ * getModelConsumption("gpt-4") // 返回 2.0
+ * getModelConsumption("unknown-model") // 返回 999
+ */
+function getModelConsumption(modelName: string): number {
+  // 将模型消耗数据转换为 Map 结构，方便快速查询
+  const consumptionMap = new Map(
+    modelConsumptionData.models.map((m) => [m.name, m.consumption]),
+  )
+  
+  // 获取消耗值，未找到则返回 "N/A"
+  const consumption = consumptionMap.get(modelName) || "N/A"
+
+  // 如果模型不在配置中，返回最大值 999
+  if (consumption === "N/A") return 999
+  
+  // 解析消耗值（格式如 "1.0x", "2.5x"）
+  const match = consumption.match(/^([\d.]+)x$/)
+  return match ? Number.parseFloat(match[1]) : 999
+}
+
+/**
+ * 检查高级交互配额使用率是否过高（超过 50%）
+ * 
+ * 当高级模型使用率超过 50% 时，系统会倾向于使用 0x 消耗的模型，
+ * 以避免快速耗尽配额
+ * 
+ * @returns true 表示使用率 >50%，false 表示使用率 ≤50% 或未初始化
+ */
+function isPremiumUsageHigh(): boolean {
+  // 如果没有高级交互数据，认为使用率不高
+  if (!state.premiumInteractions) {
+    return false
+  }
+
+  // 计算已使用百分比：100% - 剩余百分比 = 已使用百分比
+  const usagePercent = 100 - state.premiumInteractions.percent_remaining
+  return usagePercent > 50
+}
+
+/**
+ * 获取所有 0x 消耗的模型列表
+ * 
+ * 这些模型不计入高级交互配额，可以在配额紧张时优先使用
+ * 
+ * @returns 0x 消耗模型的 ID 数组
+ */
+function getZeroConsumptionModels(): string[] {
+  // 筛选出有效的可用模型（必须有上下文窗口限制配置）
+  const availableModels = state.models?.data.filter(
+    (m) => typeof m.capabilities?.limits?.max_context_window_tokens === "number",
+  )
+
+  if (!availableModels) return []
+
+  // 过滤出消耗值为 0 的模型，返回其 ID
+  return availableModels
+    .filter((m) => getModelConsumption(m.name) === 0)
+    .map((m) => m.id)
+}
+
+/**
+ * 从可用模型中查找匹配的模型
+ * 
+ * 匹配策略：
+ * 1. 标准化匹配：标准化后的模型名称匹配（下划线转连字符、版本号格式化）
+ * 2. 前缀匹配：模型名称前缀匹配
+ * 3. 基础名称匹配：忽略版本号后的基础名称匹配
+ * 
+ * 配额保护：
+ * - 当高级交互使用率 >50% 时，所有匹配仅限于 0x 消耗模型
+ * - 无匹配时降级到第一个 0x 模型
+ * 
+ * @param requestedModel - 请求的模型标识符
+ * @returns 匹配的模型 ID，未找到返回 null
+ * 
+ * @example
+ * findMatchingModel("claude-3-5-sonnet") // 返回 "claude-3.5-sonnet"
+ * findMatchingModel("gpt-4-20240101") // 返回 "gpt-4"
+ */
+export function findMatchingModel(requestedModel: string): string | null {
+  // 获取所有有效的可用模型（必须配置了上下文窗口限制）
+  const availableModels = state.models?.data.filter(
+    (m) => typeof m.capabilities?.limits?.max_context_window_tokens === "number",
+  )
+
+  // 如果没有可用模型，直接返回 null
+  if (!availableModels || availableModels.length === 0) {
+    return null
+  }
+
+  // 检查是否处于高使用率状态
+  const highUsage = isPremiumUsageHigh()
+  
+  // 如果使用率高，获取 0x 消耗模型列表用于限制匹配范围
+  const zeroConsumptionModels = highUsage ? getZeroConsumptionModels() : []
+  
+  // 提取所有可用模型的 ID
+  const allAvailableModelIds = availableModels.map((m) => m.id)
+
+  consola.debug(`正在查找匹配模型：${requestedModel}`)
+  consola.debug(`所有可用模型：${allAvailableModelIds.join(", ")}`)
+
+  // ========== 配额保护：高使用率时限制模糊匹配范围 ==========
+  let availableModelIds = allAvailableModelIds
+  if (highUsage && zeroConsumptionModels.length > 0) {
+    consola.info(
+      `⚠️  高级交互使用率 >50%，模糊匹配仅限 0x 消耗模型`,
+    )
+    availableModelIds = zeroConsumptionModels
+    consola.debug(`用于匹配的 0x 模型：${availableModelIds.join(", ")}`)
+  }
+
+  // ========== 标准化处理：统一模型名称格式 ==========
+  // 1. 转换为小写
+  // 2. 下划线转连字符（claude_3_5 -> claude-3-5）
+  // 3. 移除日期后缀（-20251001 等 8 位数字）
+  // 4. 版本号格式化（4-5 -> 4.5）
+  let normalizedRequested = requestedModel
+    .toLowerCase()
+    .replace(/_/g, "-")                 // 下划线转连字符
+    .replace(/-(\d{8})$/, "")           // 移除 -20251001 风格的日期后缀
+    .replace(/(\d)-(\d)/g, "$1.$2")     // 版本号：4-5 -> 4.5
+
+  consola.debug(`标准化后的请求模型：${normalizedRequested}`)
+
+  // ========== 策略 1：标准化后精确匹配 ==========
+  for (const availableId of availableModelIds) {
+    if (availableId.toLowerCase() === normalizedRequested) {
+      consola.info(
+        `🔄 标准化匹配成功：'${requestedModel}' -> '${availableId}'`,
+      )
+      return availableId
+    }
+  }
+
+  // ========== 策略 2：前缀匹配 ==========
+  // 检查请求的模型和可用模型是否有前缀关系
+  // 例如：claude-3.5 可以匹配 claude-3.5-sonnet-20241022
+  for (const availableId of availableModelIds) {
+    const normalizedAvailable = availableId.toLowerCase()
+
+    // 双向前缀检查：请求模型是可用模型的前缀，或可用模型是请求模型的前缀
+    if (
+      normalizedAvailable.startsWith(normalizedRequested) ||
+      normalizedRequested.startsWith(normalizedAvailable)
+    ) {
+      consola.info(
+        `🔄 前缀匹配成功：'${requestedModel}' -> '${availableId}'`,
+      )
+      return availableId
+    }
+  }
+
+  // ========== 策略 3：基础名称匹配（忽略版本号） ==========
+  // 将模型名称按 "-" 分割，比较除最后一部分外的所有部分
+  // 例如：claude-3-5-sonnet-v2 和 claude-3-5-sonnet-v1 的基础名称都是 claude-3-5-sonnet
+  const requestedParts = normalizedRequested.split("-")
+  for (const availableId of availableModelIds) {
+    const normalizedAvailable = availableId.toLowerCase()
+    const availableParts = normalizedAvailable.split("-")
+
+    // 只对至少有 3 个部分的模型名称进行基础匹配（避免过于宽泛）
+    if (requestedParts.length >= 3 && availableParts.length >= 3) {
+      // 提取基础名称（去掉最后一个部分，通常是版本号或日期）
+      const requestedBase = requestedParts.slice(0, -1).join("-")
+      const availableBase = availableParts.slice(0, -1).join("-")
+
+      if (requestedBase === availableBase) {
+        consola.info(
+          `🔄 基础名称匹配成功：'${requestedModel}' -> '${availableId}'`,
+        )
+        return availableId
+      }
+    }
+  }
+
+  // ========== 降级策略：使用率高时降级到第一个 0x 模型 ==========
+  if (highUsage && zeroConsumptionModels.length > 0) {
+    consola.warn(
+      `⚠️  未找到匹配的 0x 模型，降级到：${zeroConsumptionModels[0]}`,
+    )
+    return zeroConsumptionModels[0]
+  }
+
+  // 所有策略都失败，返回 null
+  consola.debug(`未找到匹配模型：${requestedModel}`)
+  return null
+}
+
+/**
+ * 验证并替换请求的模型
+ * 
+ * 该函数是模型匹配的主要入口点，负责：
+ * 1. 调用 findMatchingModel 查找匹配的模型
+ * 2. 验证模型是否可用
+ * 3. 返回验证结果或错误信息
+ * 
+ * @param requestedModel - 用户请求的模型标识符
+ * @returns 包含验证结果的对象
+ *   - success: true 表示验证成功，false 表示失败
+ *   - model: 匹配的模型 ID（成功时）
+ *   - error: 错误详情（失败时）
+ * 
+ * @example
+ * // 成功匹配
+ * validateAndReplaceModel("claude-3-5-sonnet")
+ * // 返回：{ success: true, model: "claude-3.5-sonnet" }
+ * 
+ * // 匹配失败
+ * validateAndReplaceModel("unknown-model")
+ * // 返回：{ success: false, error: { ... } }
+ */
+export function validateAndReplaceModel(requestedModel: string): {
+  success: boolean
+  model?: string
+  error?: {
+    message: string
+    code: string
+    param: string
+    type: string
+  }
+} {
+  // 获取所有有效的可用模型列表
+  const availableModels = state.models?.data.filter(
+    (m) => typeof m.capabilities?.limits?.max_context_window_tokens === "number",
+  )
+  const availableModelIds = availableModels?.map((m) => m.id) || []
+
+  // 尝试查找匹配的模型
+  const matchedModel = findMatchingModel(requestedModel)
+
+  // ========== 验证失败：未找到匹配的模型 ==========
+  if (!matchedModel) {
+    consola.error(`❌ 模型不可用：${requestedModel}`)
+    consola.error(`可用模型列表：${availableModelIds.join(", ")}`)
+
+    return {
+      success: false,
+      error: {
+        message: `请求的模型 '${requestedModel}' 不受支持。可用模型：${availableModelIds.join(", ")}`,
+        code: "model_not_supported",
+        param: "model",
+        type: "invalid_request_error",
+      },
+    }
+  }
+
+  // ========== 验证成功：记录结果 ==========
+  if (matchedModel !== requestedModel) {
+    // 模型被替换（通过模糊匹配找到）
+    consola.success(
+      `✓ 模型匹配并替换：${requestedModel} -> ${matchedModel}`,
+    )
+  } else {
+    // 精确匹配
+    consola.success(`✓ 模型验证通过：${matchedModel}`)
+  }
+
+  return {
+    success: true,
+    model: matchedModel,
+  }
+}
diff --git a/src/lib/refresh-usage.ts b/src/lib/refresh-usage.ts
new file mode 100644
index 00000000..eef4fddc
--- /dev/null
+++ b/src/lib/refresh-usage.ts
@@ -0,0 +1,69 @@
+import consola from "consola"
+
+import { state } from "./state"
+import { getCopilotUsage } from "~/services/github/get-copilot-usage"
+
+// Cache configuration
+const USAGE_CACHE_TTL_MS = 60 * 1000 // 1 minute cache
+let lastUsageFetchTime = 0
+let isFetching = false
+
+/**
+ * Refresh premium interactions usage information
+ * with cache to avoid excessive API calls
+ */
+export async function refreshUsage(): Promise<void> {
+  const now = Date.now()
+
+  // Check if cache is still valid
+  if (now - lastUsageFetchTime < USAGE_CACHE_TTL_MS) {
+    consola.debug(
+      `Using cached usage info (cached ${Math.floor((now - lastUsageFetchTime) / 1000)}s ago)`,
+    )
+    return
+  }
+
+  // Prevent concurrent fetches
+  if (isFetching) {
+    consola.debug("Usage fetch already in progress, skipping")
+    return
+  }
+
+  try {
+    isFetching = true
+    consola.debug("Fetching latest usage information...")
+
+    const usage = await getCopilotUsage()
+    state.premiumInteractions = usage.quota_snapshots.premium_interactions
+
+    lastUsageFetchTime = now
+
+    const usagePercent = 100 - state.premiumInteractions.percent_remaining
+    consola.debug(
+      `✓ Usage refreshed: ${usagePercent.toFixed(1)}% (${state.premiumInteractions.remaining}/${state.premiumInteractions.entitlement} remaining)`,
+    )
+  } catch (error) {
+    consola.warn("Failed to refresh usage information:", error)
+    // Continue with existing state - don't block the main flow
+  } finally {
+    isFetching = false
+  }
+}
+
+/**
+ * Force refresh usage (bypass cache)
+ */
+export async function forceRefreshUsage(): Promise<void> {
+  lastUsageFetchTime = 0
+  await refreshUsage()
+}
+
+/**
+ * Get current usage percentage
+ */
+export function getCurrentUsagePercent(): number | null {
+  if (!state.premiumInteractions) {
+    return null
+  }
+  return 100 - state.premiumInteractions.percent_remaining
+}
diff --git a/src/lib/state.ts b/src/lib/state.ts
index 5ba4dc1d..b38f83f2 100644
--- a/src/lib/state.ts
+++ b/src/lib/state.ts
@@ -1,4 +1,5 @@
 import type { ModelsResponse } from "~/services/copilot/get-models"
+import type { QuotaDetail } from "~/services/github/get-copilot-usage"
 
 export interface State {
   githubToken?: string
@@ -15,6 +16,9 @@ export interface State {
   // Rate limiting configuration
   rateLimitSeconds?: number
   lastRequestTimestamp?: number
+
+  // Usage information
+  premiumInteractions?: QuotaDetail
 }
 
 export const state: State = {
diff --git a/src/routes/chat-completions/handler.ts b/src/routes/chat-completions/handler.ts
index 04a5ae9e..b503ae5d 100644
--- a/src/routes/chat-completions/handler.ts
+++ b/src/routes/chat-completions/handler.ts
@@ -4,7 +4,9 @@ import consola from "consola"
 import { streamSSE, type SSEMessage } from "hono/streaming"
 
 import { awaitApproval } from "~/lib/approval"
+import { validateAndReplaceModel } from "~/lib/model-matcher"
 import { checkRateLimit } from "~/lib/rate-limit"
+import { refreshUsage } from "~/lib/refresh-usage"
 import { state } from "~/lib/state"
 import { getTokenCount } from "~/lib/tokenizer"
 import { isNullish } from "~/lib/utils"
@@ -15,11 +17,27 @@ import {
 } from "~/services/copilot/create-chat-completions"
 
 export async function handleCompletion(c: Context) {
+  // Refresh usage information before processing request
+  await refreshUsage()
+
   await checkRateLimit(state)
 
   let payload = await c.req.json<ChatCompletionsPayload>()
   consola.debug("Request payload:", JSON.stringify(payload).slice(-400))
 
+  // Log the requested model
+  consola.info(`Requested model: ${payload.model}`)
+
+  // Validate and potentially replace model
+  const validation = validateAndReplaceModel(payload.model)
+
+  if (!validation.success) {
+    return c.json({ error: validation.error }, 400)
+  }
+
+  // Replace model if a match was found
+  payload.model = validation.model!
+
   // Find the selected model
   const selectedModel = state.models?.data.find(
     (model) => model.id === payload.model,
diff --git a/src/routes/messages/handler.ts b/src/routes/messages/handler.ts
index 85dbf624..594aa094 100644
--- a/src/routes/messages/handler.ts
+++ b/src/routes/messages/handler.ts
@@ -4,7 +4,9 @@ import consola from "consola"
 import { streamSSE } from "hono/streaming"
 
 import { awaitApproval } from "~/lib/approval"
+import { validateAndReplaceModel } from "~/lib/model-matcher"
 import { checkRateLimit } from "~/lib/rate-limit"
+import { refreshUsage } from "~/lib/refresh-usage"
 import { state } from "~/lib/state"
 import {
   createChatCompletions,
@@ -23,6 +25,9 @@ import {
 import { translateChunkToAnthropicEvents } from "./stream-translation"
 
 export async function handleCompletion(c: Context) {
+  // Refresh usage information before processing request
+  await refreshUsage()
+
   await checkRateLimit(state)
 
   const anthropicPayload = await c.req.json<AnthropicMessagesPayload>()
@@ -34,6 +39,19 @@ export async function handleCompletion(c: Context) {
     JSON.stringify(openAIPayload),
   )
 
+  // Log the requested model
+  consola.info(`Requested model: ${openAIPayload.model}`)
+
+  // Validate and potentially replace model
+  const validation = validateAndReplaceModel(openAIPayload.model)
+
+  if (!validation.success) {
+    return c.json({ error: validation.error }, 400)
+  }
+
+  // Replace model if a match was found
+  openAIPayload.model = validation.model!
+
   if (state.manualApprove) {
     await awaitApproval()
   }
diff --git a/src/routes/models/route.ts b/src/routes/models/route.ts
index 5254e2af..a3ce0110 100644
--- a/src/routes/models/route.ts
+++ b/src/routes/models/route.ts
@@ -3,6 +3,7 @@ import { Hono } from "hono"
 import { forwardError } from "~/lib/error"
 import { state } from "~/lib/state"
 import { cacheModels } from "~/lib/utils"
+import modelConsumptionData from "~/lib/model-consumption.json"
 
 export const modelRoutes = new Hono()
 
@@ -13,15 +14,39 @@ modelRoutes.get("/", async (c) => {
       await cacheModels()
     }
 
-    const models = state.models?.data.map((model) => ({
-      id: model.id,
-      object: "model",
-      type: "model",
-      created: 0, // No date available from source
-      created_at: new Date(0).toISOString(), // No date available from source
-      owned_by: model.vendor,
-      display_name: model.name,
-    }))
+    // Create a map for quick consumption lookup
+    const consumptionMap = new Map(
+      modelConsumptionData.models.map((m) => [m.name, m.consumption]),
+    )
+
+    // Helper function to convert consumption string to number for sorting
+    const consumptionToNumber = (consumption: string): number => {
+      if (consumption === "N/A") return 999 // Put N/A at the end
+      const match = consumption.match(/^([\d.]+)x$/)
+      return match ? Number.parseFloat(match[1]) : 999
+    }
+
+    // Filter to only include models with context window information (Available models)
+    const models = state.models?.data
+      .filter((model) => {
+        const maxTokens = model.capabilities?.limits?.max_context_window_tokens
+        return typeof maxTokens === "number"
+      })
+      .map((model) => ({
+        model,
+        consumption: consumptionMap.get(model.name) || "N/A",
+      }))
+      .sort((a, b) => consumptionToNumber(a.consumption) - consumptionToNumber(b.consumption))
+      .map((item) => ({
+        id: item.model.id,
+        object: "model",
+        type: "model",
+        created: 0, // No date available from source
+        created_at: new Date(0).toISOString(), // No date available from source
+        owned_by: item.model.vendor,
+        display_name: item.model.name,
+        max_context_length: item.model.capabilities?.limits?.max_context_window_tokens,
+      }))
 
     return c.json({
       object: "list",
diff --git a/src/server.ts b/src/server.ts
index 462a278f..ce897ed6 100644
--- a/src/server.ts
+++ b/src/server.ts
@@ -27,5 +27,10 @@ server.route("/v1/chat/completions", completionRoutes)
 server.route("/v1/models", modelRoutes)
 server.route("/v1/embeddings", embeddingRoutes)
 
+// Compatibility with tools that expect api/v0/ prefix
+server.route("/api/v0/models", modelRoutes)
+server.route("/api/v0/chat/completions", completionRoutes)
+server.route("/api/v0/embeddings", embeddingRoutes)
+
 // Anthropic compatible endpoints
 server.route("/v1/messages", messageRoutes)
diff --git a/src/services/copilot/create-chat-completions.ts b/src/services/copilot/create-chat-completions.ts
index 8534151d..94966946 100644
--- a/src/services/copilot/create-chat-completions.ts
+++ b/src/services/copilot/create-chat-completions.ts
@@ -35,7 +35,29 @@ export const createChatCompletions = async (
   })
 
   if (!response.ok) {
-    consola.error("Failed to create chat completions", response)
+    const errorBody = await response.text()
+    consola.error(`Failed to create chat completions for model: ${payload.model}`)
+    consola.error(`Response status: ${response.status} ${response.statusText}`)
+    consola.error(`Response body: ${errorBody}`)
+    
+    // Try to parse error details
+    try {
+      const errorJson = JSON.parse(errorBody)
+      if (errorJson.error?.message) {
+        consola.error(`Error message: ${errorJson.error.message}`)
+        
+        // If model not supported, list available models
+        if (errorJson.error.code === "model_not_supported") {
+          const availableModels = state.models?.data
+            .filter((m) => typeof m.capabilities?.limits?.max_context_window_tokens === "number")
+            .map((m) => m.id)
+          consola.error(`Available models: ${availableModels?.join(", ")}`)
+        }
+      }
+    } catch {
+      // If parsing fails, we already logged the raw body
+    }
+    
     throw new HTTPError("Failed to create chat completions", response)
   }
 
diff --git a/src/services/copilot/get-models.ts b/src/services/copilot/get-models.ts
index 3cfa30af..42d96b3f 100644
--- a/src/services/copilot/get-models.ts
+++ b/src/services/copilot/get-models.ts
@@ -9,7 +9,12 @@ export const getModels = async () => {
 
   if (!response.ok) throw new HTTPError("Failed to get models", response)
 
-  return (await response.json()) as ModelsResponse
+  const result = await response.json() as ModelsResponse
+  result.data = result.data.filter(
+    (model: any) =>
+      model.model_picker_category !== undefined && model.model_picker_enabled === true
+  )
+  return result
 }
 
 export interface ModelsResponse {
@@ -48,6 +53,7 @@ export interface Model {
   preview: boolean
   vendor: string
   version: string
+  model_picker_category?: string
   policy?: {
     state: string
     terms: string
diff --git a/src/start.ts b/src/start.ts
index 14abbbdf..bc03521b 100644
--- a/src/start.ts
+++ b/src/start.ts
@@ -13,6 +13,7 @@ import { state } from "./lib/state"
 import { setupCopilotToken, setupGitHubToken } from "./lib/token"
 import { cacheModels, cacheVSCodeVersion } from "./lib/utils"
 import { server } from "./server"
+import modelConsumptionData from "./lib/model-consumption.json"
 
 interface RunServerOptions {
   port: number
@@ -60,8 +61,72 @@ export async function runServer(options: RunServerOptions): Promise<void> {
   await setupCopilotToken()
   await cacheModels()
 
+  // Initial fetch of usage information for premium interactions tracking
+  const { forceRefreshUsage, getCurrentUsagePercent } = await import(
+    "./lib/refresh-usage"
+  )
+  await forceRefreshUsage()
+
+  const usagePercent = getCurrentUsagePercent()
+  if (usagePercent !== null && state.premiumInteractions) {
+    consola.info(
+      `Premium interactions usage: ${usagePercent.toFixed(1)}% (${state.premiumInteractions.remaining}/${state.premiumInteractions.entitlement} remaining)`,
+    )
+  }
+
+  // consola.info(
+  //   `Full Model Info:\n${
+  //     state.models?.data
+  //       ?.filter(model => model.model_picker_enabled === true)
+  //       .map(model => {
+  //         const { 
+  //           capabilities, 
+  //           policy, 
+  //           vendor, 
+  //           preview, 
+  //           model_picker_enabled,
+  //           object,
+  //           ...rest } = model;
+  //         let fullInfo = JSON.stringify(rest, null, 2);
+  //         return `- ${model.id}\n${fullInfo}`;
+  //       }).join("\n")
+  //   }`
+  // )
+
+  // Create a map for quick consumption lookup
+  const consumptionMap = new Map(
+    modelConsumptionData.models.map(m => [m.name, m.consumption])
+  );
+
+  // Helper function to convert consumption string to number for sorting
+  const consumptionToNumber = (consumption: string): number => {
+    if (consumption === "N/A") return 999; // Put N/A at the end
+    const match = consumption.match(/^([\d.]+)x$/);
+    return match ? Number.parseFloat(match[1]) : 999;
+  };
+
   consola.info(
-    `Available models: \n${state.models?.data.map((model) => `- ${model.id}`).join("\n")}`,
+    `Available models:\n${
+      state.models?.data
+        .map(model => {
+          let maxTokens = model.capabilities?.limits?.max_context_window_tokens;
+          let maxTokensStr = "N/A";
+          if (typeof maxTokens === "number") {
+            maxTokensStr = maxTokens >= 1000 ? `${maxTokens / 1000}k` : `${maxTokens}`;
+          }
+          const consumption = consumptionMap.get(model.name) || "N/A";
+          return { model, maxTokensStr, consumption };
+        })
+        .filter(item => item.maxTokensStr !== "N/A")
+        .sort((a, b) => consumptionToNumber(a.consumption) - consumptionToNumber(b.consumption))
+        .map((item, i) => {
+          const consumptionStr = `(${item.consumption})`.padEnd(8, " ");
+          const idStr = item.model.id.padEnd(24, " ");
+          const nameStr = item.model.name.padEnd(32, " ");
+          const contextStr = `context: ${item.maxTokensStr}`;
+          return `${i + 1}.\t${consumptionStr}${idStr}\t${nameStr}\t${contextStr}`;
+        }).join("\n")
+    }`
   )
 
   const serverUrl = `http://localhost:${options.port}`
diff --git a/start.bat b/start.bat
index 1a0f8cb8..d101ebd8 100644
--- a/start.bat
+++ b/start.bat
@@ -15,6 +15,6 @@ echo The usage viewer page will open automatically after the server starts
 echo.
 
 start "" "https://ericc-ch.github.io/copilot-api?endpoint=http://localhost:4141/usage"
-bun run dev
+bun run --watch ./src/main.ts start
 
 pause
diff --git a/tsconfig.json b/tsconfig.json
index 8ff821ca..5f43fef1 100644
--- a/tsconfig.json
+++ b/tsconfig.json
@@ -6,6 +6,7 @@
     "skipLibCheck": true,
 
     "allowJs": true,
+    "resolveJsonModule": true,
     "moduleResolution": "Bundler",
     "moduleDetection": "force",
     "erasableSyntaxOnly": true,