mohdel 1.1.0 → 1.3.0

package/src/lib/provider-info.js

@@ -86,6 +86,13 @@ const PROVIDER_INFO = {
     hint: 'Create an API key in the MiMo open platform console',
     free: false
   },
+  cohere: {
+    label: 'Cohere',
+    description: 'Embeddings and reranking for retrieval. No chat models through mohdel.',
+    url: 'https://dashboard.cohere.com/api-keys',
+    hint: 'Create an API key in the Cohere dashboard under API Keys',
+    free: true
+  },
   qwen: {
     label: 'Qwen Cloud',
     description: 'Qwen — reasoning, coding, long context. Free quota for new users.',

package/src/lib/providers.js

@@ -86,6 +86,18 @@ const providers = {
     contextSemantics: 'shared',
     outputCapStrategy: 'accept'
   },
+  cohere: {
+    sdk: 'cohere',
+    api: 'embeddings',
+    apiKeyEnv: 'COHERE_API_SK',
+    baseURL: 'https://api.cohere.com/v2',
+    createConfiguration: apiKey => ({ apiKey }),
+    references: {
+      pricing: 'https://cohere.com/pricing',
+      models: 'https://docs.cohere.com/docs/models',
+      rateLimits: 'https://docs.cohere.com/docs/rate-limits'
+    }
+  },
   mistral: {
     sdk: 'openai',
     api: 'chatCompletions',

package/src/lib/schema.js

@@ -58,10 +58,18 @@ const fieldDefs = {
   imageEndpoint: { type: 'string' },
   imageDefaultSize: { type: 'string' },
   transcriptionPrice: { type: 'number' },
+  embeddingPrice: { type: 'number' },
+  dimensions: { type: 'number' },
+  dimensionsSelectable: { type: 'boolean' },
+  maxBatch: { type: 'number' },
+  maxInputTokens: { type: 'number' },
+  inputTypes: { type: 'object' },
+  defaultInputType: { type: 'string' },
   deprecated: { type: 'string' },
   suspended: { type: 'string' },
   rpmLimit: { type: 'number' },
   tpmLimit: { type: 'number' },
+  inpmLimit: { type: 'number' },
   rateLimitScope: { type: 'string', validate: (v) => ['model', 'provider'].includes(v) ? null : 'must be "model" or "provider"' },
   outputCapStrategy: { type: 'string', validate: (v) => ['error', 'accept'].includes(v) ? null : "must be 'error' or 'accept'" },
   supportsTools: { type: 'boolean' },
@@ -77,6 +85,7 @@ const COMPUTED_FIELDS = new Set(['upstreamIds'])
 const TYPE_CHECKERS = {
   string: (v) => typeof v === 'string',
   number: (v) => typeof v === 'number',
+  boolean: (v) => typeof v === 'boolean',
   array: (v) => Array.isArray(v),
   object: (v) => typeof v === 'object' && v !== null && !Array.isArray(v)
 }

package/src/lib/rate-limiter.js

@@ -1,50 +0,0 @@
-// Lightweight per-minute rate limiter.
-// Tracks RPM and TPM with minute-bucket granularity.
-// Throttles (delays) rather than rejects — returns ms to wait.
-
-const createRateLimiter = () => {
-  // key → { count, tokens, minute }
-  const buckets = new Map()
-
-  const currentMinute = () => Math.floor(Date.now() / 60000)
-
-  const getBucket = (key) => {
-    const minute = currentMinute()
-    const bucket = buckets.get(key)
-    if (bucket && bucket.minute === minute) return bucket
-    const fresh = { count: 0, tokens: 0, minute }
-    buckets.set(key, fresh)
-    return fresh
-  }
-
-  const msUntilNextMinute = (minute) => Math.max(0, (minute + 1) * 60000 - Date.now())
-
-  // Returns ms to wait before sending (0 = go ahead)
-  const check = (key, { rpmLimit, tpmLimit } = {}) => {
-    if (!rpmLimit && !tpmLimit) return 0
-    const bucket = getBucket(key)
-    if (rpmLimit && bucket.count >= rpmLimit) {
-      return msUntilNextMinute(bucket.minute)
-    }
-    if (tpmLimit && bucket.tokens >= tpmLimit) {
-      return msUntilNextMinute(bucket.minute)
-    }
-    return 0
-  }
-
-  // Record a request count (call before sending — RPM tracking)
-  const recordRequest = (key) => {
-    const bucket = getBucket(key)
-    bucket.count++
-  }
-
-  // Record token usage (call after response — TPM tracking)
-  const recordTokens = (key, tokens) => {
-    const bucket = getBucket(key)
-    bucket.tokens += tokens
-  }
-
-  return { check, recordRequest, recordTokens }
-}
-
-export default createRateLimiter