ai-sdk-rate-limiter 0.1.0 → 0.3.0

package/README.md

@@ -77,7 +77,7 @@ const limiter = createRateLimiter({
       daily:   50,
       monthly: 500,
     },
-    onExceeded: 'throw', // or 'queue' — wait until the period resets
+    onExceeded: 'throw', // 'throw' | 'queue' | 'fallback'
   },
 
   // Queue behavior
@@ -169,6 +169,49 @@ Costs are based on **actual token counts** from API responses — not estimates.
 
 ---
 
+## Budget fallback routing
+
+When a budget limit is hit, you can transparently reroute to a cheaper model instead of throwing an error. Pass a `fallback` option to `wrap()`:
+
+```typescript
+const limiter = createRateLimiter({
+  cost: {
+    budget: { daily: 10 },
+    onExceeded: 'fallback',  // reroute to fallback instead of throwing
+  },
+  on: {
+    budgetHit: ({ model, currentCostUsd, limitUsd, period }) =>
+      console.warn(`${model} ${period} budget hit ($${currentCostUsd} of $${limitUsd})`),
+  },
+})
+
+const model = limiter.wrap(
+  openai('gpt-4o'),                     // primary model
+  { fallback: openai('gpt-4o-mini') },  // used when budget is exceeded
+)
+
+// Under budget  → uses gpt-4o normally
+// Over $10/day  → silently switches to gpt-4o-mini, no code changes needed
+const result = await generateText({ model, prompt })
+```
+
+**How it works:**
+1. The budget is checked before every request against total rolling spend
+2. When exceeded, `BudgetExceededError` is caught inside `wrap()` before it reaches your code
+3. The request is re-executed against the fallback model, bypassing the budget pre-check
+4. Fallback usage is tracked under the fallback model's ID in `getCostReport()`
+
+**Behavior matrix:**
+
+| `onExceeded` | `fallback` configured | Outcome |
+|---|---|---|
+| `'throw'` | any | Throws `BudgetExceededError` |
+| `'fallback'` | yes | Transparently uses fallback model |
+| `'fallback'` | no | Throws `BudgetExceededError` |
+| `'queue'` | any | Queues until period resets |
+
+---
+
 ## Backpressure — know before you send
 
 Check estimated wait time before committing to a request. Useful for showing loading states or shedding load gracefully.
@@ -209,7 +252,7 @@ limiter.off('queued', handler)
 | `dequeued` | Request leaves the queue | `model`, `waitedMs`, `priority` |
 | `retrying` | A failed request is about to retry | `model`, `attempt`, `maxAttempts`, `delayMs`, `error` |
 | `rateLimited` | Limit hit (local or remote 429) | `model`, `source`, `limitType`, `resetAt` |
-| `budgetHit` | Cost budget exceeded | `model`, `currentCostUsd`, `limitUsd`, `period` |
+| `budgetHit` | Cost budget exceeded | `model`, `currentCostUsd`, `limitUsd`, `period`, `usingFallback` |
 | `dropped` | Request rejected (queue full or timeout) | `model`, `reason` |
 | `completed` | Request finished successfully | `model`, `inputTokens`, `outputTokens`, `costUsd`, `latencyMs` |
 
@@ -304,7 +347,7 @@ const model = req.user.plan === 'paid'
 
 ## Built-in model registry
 
-Limits and pricing are built-in for every major model. These defaults are Tier 1 (most conservative) — override with your actual tier limits.
+Limits and pricing are built-in for every major model across 6 providers. Defaults are conservative (free/Tier 1) — override with your actual plan limits.
 
 **OpenAI**
 
@@ -332,13 +375,54 @@ Limits and pricing are built-in for every major model. These defaults are Tier 1
 | gemini-1.5-pro | 2 | 32,000 | $1.25 | $5.00 |
 | gemini-1.5-flash | 15 | 1,000,000 | $0.075 | $0.30 |
 
+**Groq** (free tier defaults — on-demand tier is 6,000 RPM / 200k TPM)
+
+| Model | RPM | ITPM | Input $/M | Output $/M |
+|---|---|---|---|---|
+| llama-3.3-70b-versatile | 30 | 6,000 | $0.59 | $0.79 |
+| llama-3.1-8b-instant | 30 | 20,000 | $0.05 | $0.08 |
+| mixtral-8x7b-32768 | 30 | 5,000 | $0.24 | $0.24 |
+| gemma2-9b-it | 30 | 15,000 | $0.20 | $0.20 |
+| deepseek-r1-distill-llama-70b | 30 | 6,000 | $0.75 | $0.99 |
+
+**Mistral**
+
+| Model | RPM | ITPM | Input $/M | Output $/M |
+|---|---|---|---|---|
+| mistral-large-latest | 500 | 100,000 | $2.00 | $6.00 |
+| mistral-small-latest | 500 | 100,000 | $0.10 | $0.30 |
+| codestral-latest | 500 | 100,000 | $0.30 | $0.90 |
+| open-mistral-nemo | 500 | 100,000 | $0.15 | $0.15 |
+| pixtral-large-latest | 500 | 100,000 | $2.00 | $6.00 |
+
+**Cohere** (trial tier defaults — production tier is 10,000+ RPM)
+
+| Model | RPM | ITPM | Input $/M | Output $/M |
+|---|---|---|---|---|
+| command-r-plus | 20 | 100,000 | $2.50 | $10.00 |
+| command-r | 20 | 100,000 | $0.15 | $0.60 |
+| command | 20 | 100,000 | $0.50 | $1.50 |
+| command-light | 20 | 100,000 | $0.15 | $0.60 |
+
 Unknown models fall back to 60 RPM / 100k ITPM with no cost tracking. You can inspect or extend the registry:
 
 ```typescript
-import { OPENAI_MODELS, ANTHROPIC_MODELS, resolveModelLimits, isKnownModel } from 'ai-sdk-rate-limiter'
+import {
+  OPENAI_MODELS,
+  ANTHROPIC_MODELS,
+  GOOGLE_MODELS,
+  GROQ_MODELS,
+  MISTRAL_MODELS,
+  COHERE_MODELS,
+  resolveModelLimits,
+  isKnownModel,
+} from 'ai-sdk-rate-limiter'
+
+console.log(GROQ_MODELS['llama-3.3-70b-versatile'])
+// { rpm: 30, itpm: 6000, rpd: 1000, inputPricePerMillion: 0.59, ... }
 
-console.log(OPENAI_MODELS['gpt-4o'])
-// { rpm: 500, itpm: 30000, otpm: 30000, inputPricePerMillion: 2.5, ... }
+console.log(isKnownModel('llama-3.3-70b-versatile', 'groq'))
+// true
 
 console.log(isKnownModel('my-fine-tune', 'openai'))
 // false → will use fallback limits

package/dist/index.cjs

@@ -317,7 +317,7 @@ var CostTracker = class {
     ];
     for (const { limit, current, period } of checks) {
       if (limit !== void 0 && current + estimatedCostUsd > limit) {
-        if (onExceeded === "throw") {
+        if (onExceeded === "throw" || onExceeded === "fallback") {
           throw new BudgetExceededError(model, current, limit, period);
         }
         return false;
@@ -873,6 +873,341 @@ var GOOGLE_MODELS = {
   }
 };
 
+// src/registry/groq.ts
+var GROQ_MODELS = {
+  // -------------------------------------------------------------------------
+  // Llama 3.3 family
+  // -------------------------------------------------------------------------
+  "llama-3.3-70b-versatile": {
+    rpm: 30,
+    itpm: 6e3,
+    otpm: 6e3,
+    rpd: 1e3,
+    inputPricePerMillion: 0.59,
+    outputPricePerMillion: 0.79
+  },
+  "llama-3.3-70b-specdec": {
+    rpm: 30,
+    itpm: 6e3,
+    otpm: 6e3,
+    rpd: 1e3,
+    inputPricePerMillion: 0.59,
+    outputPricePerMillion: 0.99
+  },
+  // -------------------------------------------------------------------------
+  // Llama 3.1 family
+  // -------------------------------------------------------------------------
+  "llama-3.1-8b-instant": {
+    rpm: 30,
+    itpm: 2e4,
+    otpm: 2e4,
+    rpd: 14400,
+    inputPricePerMillion: 0.05,
+    outputPricePerMillion: 0.08
+  },
+  "llama-3.1-70b-versatile": {
+    rpm: 30,
+    itpm: 6e3,
+    otpm: 6e3,
+    rpd: 1e3,
+    inputPricePerMillion: 0.59,
+    outputPricePerMillion: 0.79
+  },
+  // -------------------------------------------------------------------------
+  // Llama 3 family
+  // -------------------------------------------------------------------------
+  "llama3-70b-8192": {
+    rpm: 30,
+    itpm: 6e3,
+    otpm: 6e3,
+    rpd: 14400,
+    inputPricePerMillion: 0.59,
+    outputPricePerMillion: 0.79
+  },
+  "llama3-8b-8192": {
+    rpm: 30,
+    itpm: 3e4,
+    otpm: 3e4,
+    rpd: 14400,
+    inputPricePerMillion: 0.05,
+    outputPricePerMillion: 0.08
+  },
+  "llama-guard-3-8b": {
+    rpm: 30,
+    itpm: 15e3,
+    otpm: 15e3,
+    rpd: 14400,
+    inputPricePerMillion: 0.2,
+    outputPricePerMillion: 0.2
+  },
+  // -------------------------------------------------------------------------
+  // Mixtral family
+  // -------------------------------------------------------------------------
+  "mixtral-8x7b-32768": {
+    rpm: 30,
+    itpm: 5e3,
+    otpm: 5e3,
+    rpd: 14400,
+    inputPricePerMillion: 0.24,
+    outputPricePerMillion: 0.24
+  },
+  // -------------------------------------------------------------------------
+  // Gemma family
+  // -------------------------------------------------------------------------
+  "gemma2-9b-it": {
+    rpm: 30,
+    itpm: 15e3,
+    otpm: 15e3,
+    rpd: 14400,
+    inputPricePerMillion: 0.2,
+    outputPricePerMillion: 0.2
+  },
+  "gemma-7b-it": {
+    rpm: 30,
+    itpm: 15e3,
+    otpm: 15e3,
+    rpd: 14400,
+    inputPricePerMillion: 0.07,
+    outputPricePerMillion: 0.07
+  },
+  // -------------------------------------------------------------------------
+  // Deepseek family
+  // -------------------------------------------------------------------------
+  "deepseek-r1-distill-llama-70b": {
+    rpm: 30,
+    itpm: 6e3,
+    otpm: 6e3,
+    rpd: 1e3,
+    inputPricePerMillion: 0.75,
+    outputPricePerMillion: 0.99
+  },
+  "deepseek-r1-distill-qwen-32b": {
+    rpm: 30,
+    itpm: 6e3,
+    otpm: 6e3,
+    rpd: 1e3,
+    inputPricePerMillion: 0.69,
+    outputPricePerMillion: 0.69
+  }
+};
+
+// src/registry/mistral.ts
+var MISTRAL_MODELS = {
+  // -------------------------------------------------------------------------
+  // Mistral Large — frontier model
+  // -------------------------------------------------------------------------
+  "mistral-large-latest": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2,
+    outputPricePerMillion: 6
+  },
+  "mistral-large-2411": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2,
+    outputPricePerMillion: 6
+  },
+  "mistral-large-2407": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2,
+    outputPricePerMillion: 6
+  },
+  // -------------------------------------------------------------------------
+  // Mistral Small — efficient, low-cost
+  // -------------------------------------------------------------------------
+  "mistral-small-latest": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.1,
+    outputPricePerMillion: 0.3
+  },
+  "mistral-small-2409": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.1,
+    outputPricePerMillion: 0.3
+  },
+  // -------------------------------------------------------------------------
+  // Pixtral Large — multimodal
+  // -------------------------------------------------------------------------
+  "pixtral-large-latest": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2,
+    outputPricePerMillion: 6
+  },
+  "pixtral-large-2411": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2,
+    outputPricePerMillion: 6
+  },
+  "pixtral-12b": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.15,
+    outputPricePerMillion: 0.15
+  },
+  "pixtral-12b-2409": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.15,
+    outputPricePerMillion: 0.15
+  },
+  // -------------------------------------------------------------------------
+  // Codestral — code-optimized
+  // -------------------------------------------------------------------------
+  "codestral-latest": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.3,
+    outputPricePerMillion: 0.9
+  },
+  "codestral-2501": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.3,
+    outputPricePerMillion: 0.9
+  },
+  // -------------------------------------------------------------------------
+  // Open models (free / self-hosted weights available)
+  // -------------------------------------------------------------------------
+  "open-mistral-nemo": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.15,
+    outputPricePerMillion: 0.15
+  },
+  "open-mixtral-8x22b": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2,
+    outputPricePerMillion: 6
+  },
+  "open-mixtral-8x7b": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.7,
+    outputPricePerMillion: 0.7
+  },
+  "open-mistral-7b": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.25,
+    outputPricePerMillion: 0.25
+  },
+  // -------------------------------------------------------------------------
+  // Mistral Embed — embedding only (no RPM-based generation limits)
+  // -------------------------------------------------------------------------
+  "mistral-embed": {
+    rpm: 500,
+    itpm: 1e5,
+    otpm: 0,
+    inputPricePerMillion: 0.1,
+    outputPricePerMillion: 0
+  }
+};
+
+// src/registry/cohere.ts
+var COHERE_MODELS = {
+  // -------------------------------------------------------------------------
+  // Command R+ — highest capability
+  // -------------------------------------------------------------------------
+  "command-r-plus": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2.5,
+    outputPricePerMillion: 10
+  },
+  "command-r-plus-08-2024": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2.5,
+    outputPricePerMillion: 10
+  },
+  "command-r-plus-04-2024": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 2.5,
+    outputPricePerMillion: 10
+  },
+  // -------------------------------------------------------------------------
+  // Command R — balanced, RAG-optimized
+  // -------------------------------------------------------------------------
+  "command-r": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.15,
+    outputPricePerMillion: 0.6
+  },
+  "command-r-08-2024": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.15,
+    outputPricePerMillion: 0.6
+  },
+  "command-r-03-2024": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.15,
+    outputPricePerMillion: 0.6
+  },
+  // -------------------------------------------------------------------------
+  // Command — legacy general-purpose
+  // -------------------------------------------------------------------------
+  "command": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.5,
+    outputPricePerMillion: 1.5
+  },
+  "command-nightly": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.5,
+    outputPricePerMillion: 1.5
+  },
+  "command-light": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.15,
+    outputPricePerMillion: 0.6
+  },
+  "command-light-nightly": {
+    rpm: 20,
+    itpm: 1e5,
+    otpm: 1e5,
+    inputPricePerMillion: 0.15,
+    outputPricePerMillion: 0.6
+  }
+};
+
 // src/registry/index.ts
 var FALLBACK_LIMITS = {
   rpm: 60,
@@ -919,7 +1254,22 @@ function getFromRegistry(modelId, provider) {
     const stripped = modelId.replace(/^(google|vertex)\//, "");
     if (GOOGLE_MODELS[stripped]) return GOOGLE_MODELS[stripped];
   }
-  return OPENAI_MODELS[modelId] ?? ANTHROPIC_MODELS[modelId] ?? GOOGLE_MODELS[modelId];
+  if (provider === "groq") {
+    if (GROQ_MODELS[modelId]) return GROQ_MODELS[modelId];
+    const stripped = modelId.replace(/^groq\//, "");
+    if (GROQ_MODELS[stripped]) return GROQ_MODELS[stripped];
+  }
+  if (provider === "mistral") {
+    if (MISTRAL_MODELS[modelId]) return MISTRAL_MODELS[modelId];
+    const stripped = modelId.replace(/^mistral\//, "");
+    if (MISTRAL_MODELS[stripped]) return MISTRAL_MODELS[stripped];
+  }
+  if (provider === "cohere") {
+    if (COHERE_MODELS[modelId]) return COHERE_MODELS[modelId];
+    const stripped = modelId.replace(/^cohere\//, "");
+    if (COHERE_MODELS[stripped]) return COHERE_MODELS[stripped];
+  }
+  return OPENAI_MODELS[modelId] ?? ANTHROPIC_MODELS[modelId] ?? GOOGLE_MODELS[modelId] ?? GROQ_MODELS[modelId] ?? MISTRAL_MODELS[modelId] ?? COHERE_MODELS[modelId];
 }
 function isKnownModel(modelId, provider) {
   return getFromRegistry(modelId, normalizeProvider(provider)) !== void 0;
@@ -970,7 +1320,7 @@ var Pipeline = class {
     const estimatedInput = estimateInputTokens(prompt);
     const startMs = Date.now();
     const key = `${provider}:${modelId}`;
-    if (this.config.cost?.budget) {
+    if (this.config.cost?.budget && !opts.skipBudgetCheck) {
       const estimatedCost = this.costTracker.estimateCost(
         estimatedInput,
         500,
@@ -978,12 +1328,26 @@ var Pipeline = class {
         limits.inputPricePerMillion,
         limits.outputPricePerMillion
       );
-      this.costTracker.checkBudget(
-        modelId,
-        estimatedCost,
-        this.config.cost.budget,
-        this.config.cost.onExceeded ?? "throw"
-      );
+      try {
+        this.costTracker.checkBudget(
+          modelId,
+          estimatedCost,
+          this.config.cost.budget,
+          this.config.cost.onExceeded ?? "throw"
+        );
+      } catch (err) {
+        if (err instanceof BudgetExceededError) {
+          this.emitter.emit("budgetHit", {
+            model: err.model,
+            provider,
+            currentCostUsd: err.currentCostUsd,
+            limitUsd: err.limitUsd,
+            period: err.period,
+            usingFallback: false
+          });
+        }
+        throw err;
+      }
     }
     await this.engine.acquire(key, {
       limits,
@@ -1122,7 +1486,8 @@ function getPerRequestOptions(params, queueTimeout) {
   return {
     priority: raw?.priority ?? "normal",
     timeoutMs: raw?.timeout ?? queueTimeout,
-    metadata: raw?.metadata ?? {}
+    metadata: raw?.metadata ?? {},
+    skipBudgetCheck: raw?._skipBudgetCheck ?? false
   };
 }
 function extractTokenUsage(usage) {
@@ -1138,7 +1503,7 @@ function createMiddleware(pipeline, queueTimeout) {
     // wrapGenerate — non-streaming
     // -----------------------------------------------------------------------
     async wrapGenerate({ doGenerate, params, model }) {
-      const { priority, timeoutMs } = getPerRequestOptions(params, queueTimeout);
+      const { priority, timeoutMs, skipBudgetCheck } = getPerRequestOptions(params, queueTimeout);
       const modelId = model.modelId;
       const provider = model.provider;
       const startMs = Date.now();
@@ -1151,6 +1516,7 @@ function createMiddleware(pipeline, queueTimeout) {
           streaming: false,
           priority,
           timeoutMs,
+          skipBudgetCheck,
           onUsage: () => {
           }
         }
@@ -1165,7 +1531,7 @@ function createMiddleware(pipeline, queueTimeout) {
     // wrapStream — streaming
     // -----------------------------------------------------------------------
     async wrapStream({ doStream, params, model }) {
-      const { priority, timeoutMs } = getPerRequestOptions(params, queueTimeout);
+      const { priority, timeoutMs, skipBudgetCheck } = getPerRequestOptions(params, queueTimeout);
       const modelId = model.modelId;
       const provider = model.provider;
       const startMs = Date.now();
@@ -1178,6 +1544,7 @@ function createMiddleware(pipeline, queueTimeout) {
           streaming: true,
           priority,
           timeoutMs,
+          skipBudgetCheck,
           onUsage: () => {
           }
         }
@@ -1204,26 +1571,71 @@ function createMiddleware(pipeline, queueTimeout) {
 function wrapModel(model, middleware, overrides) {
   const providerId = overrides?.providerId ?? model.provider;
   const modelId = overrides?.modelId ?? model.modelId;
+  const fallbackModel = overrides?.fallback;
   return {
     specificationVersion: "v4",
     provider: providerId,
     modelId,
     supportedUrls: model["supportedUrls"],
     async doGenerate(params) {
-      return middleware.wrapGenerate({
-        doGenerate: () => model.doGenerate(params),
-        doStream: () => model.doStream(params),
-        params,
-        model
-      });
+      try {
+        return await middleware.wrapGenerate({
+          doGenerate: () => model.doGenerate(params),
+          doStream: () => model.doStream(params),
+          params,
+          model
+        });
+      } catch (err) {
+        if (err instanceof BudgetExceededError && fallbackModel) {
+          const fallbackParams = {
+            ...params,
+            providerOptions: {
+              ...params.providerOptions,
+              rateLimiter: {
+                ...params.providerOptions?.["rateLimiter"] ?? {},
+                _skipBudgetCheck: true
+              }
+            }
+          };
+          return middleware.wrapGenerate({
+            doGenerate: () => fallbackModel.doGenerate(fallbackParams),
+            doStream: () => fallbackModel.doStream(fallbackParams),
+            params: fallbackParams,
+            model: fallbackModel
+          });
+        }
+        throw err;
+      }
     },
     async doStream(params) {
-      return middleware.wrapStream({
-        doGenerate: () => model.doGenerate(params),
-        doStream: () => model.doStream(params),
-        params,
-        model
-      });
+      try {
+        return await middleware.wrapStream({
+          doGenerate: () => model.doGenerate(params),
+          doStream: () => model.doStream(params),
+          params,
+          model
+        });
+      } catch (err) {
+        if (err instanceof BudgetExceededError && fallbackModel) {
+          const fallbackParams = {
+            ...params,
+            providerOptions: {
+              ...params.providerOptions,
+              rateLimiter: {
+                ...params.providerOptions?.["rateLimiter"] ?? {},
+                _skipBudgetCheck: true
+              }
+            }
+          };
+          return middleware.wrapStream({
+            doGenerate: () => fallbackModel.doGenerate(fallbackParams),
+            doStream: () => fallbackModel.doStream(fallbackParams),
+            params: fallbackParams,
+            model: fallbackModel
+          });
+        }
+        throw err;
+      }
     }
   };
 }
@@ -1260,7 +1672,10 @@ function createRateLimiter(config = {}) {
 
 exports.ANTHROPIC_MODELS = ANTHROPIC_MODELS;
 exports.BudgetExceededError = BudgetExceededError;
+exports.COHERE_MODELS = COHERE_MODELS;
 exports.GOOGLE_MODELS = GOOGLE_MODELS;
+exports.GROQ_MODELS = GROQ_MODELS;
+exports.MISTRAL_MODELS = MISTRAL_MODELS;
 exports.OPENAI_MODELS = OPENAI_MODELS;
 exports.QueueFullError = QueueFullError;
 exports.QueueTimeoutError = QueueTimeoutError;