mohdel 1.0.3 → 1.1.0

package/README.md

@@ -452,17 +452,17 @@ What each provider supports through mohdel's unified interface:
 | Anthropic | Yes | Yes | Yes | No | Yes (adaptive / budget) | `identifier` → `metadata.user_id` |
 | OpenAI | Yes | Yes | Yes | No | Yes (o-series) | GPT-5 verbosity via `outputStyle` |
 | Gemini | Yes | Yes | Yes | Yes | Yes (`thinkingLevel` / `thinkingBudget`) | Auto-uploads large videos; content-hashed cache |
-| Cerebras | No | Yes | Yes | No | Yes (`reasoning_effort` or zai `disable_reasoning`) | Non-streaming chat completions |
-| Groq | No | Yes | Yes | No | No | Non-streaming; shared chat-completions path |
+| Cerebras | Yes | Yes | Yes | No | Yes (`reasoning_effort` or zai `disable_reasoning`) | Shared chat-completions path |
+| Groq | Yes | Yes | Yes | No | No | Shared chat-completions path |
 | xAI | Yes | Yes | Yes | No | Auto | OpenAI Responses API over `api.x.ai/v1` |
-| DeepSeek | No | Yes | Yes | No | No | DSML tool-call fallback when model emits tags in content |
+| DeepSeek | No | Yes | Yes | No | No | Non-streaming: the DSML tool-call fallback is only parsed off a complete response |
 | Fireworks | Yes | Yes | Yes | No | Yes (`reasoning_effort`) | OpenAI SDK + `baseURL`; model id auto-prefixed |
-| Mistral | No | Yes | Yes | No | No | `tool_choice: "any"` = required |
-| Qwen Cloud | No | Yes | No | No | Yes (`enable_thinking` + `thinking_budget`) | Alibaba DashScope intl; hybrid models think by default — effort `none` sends explicit off |
-| Xiaomi | No | Yes | Yes | No | Auto | MiMo; shared chat-completions path, `reasoning_content` captured |
+| Mistral | Yes | Yes | Yes | No | No | `tool_choice: "any"` = required |
+| Qwen Cloud | Yes | Yes | No | No | Yes (`enable_thinking` + `thinking_budget`) | Alibaba DashScope intl; hybrid models think by default — effort `none` sends explicit off |
+| Xiaomi | Yes | Yes | Yes | No | Auto | MiMo; shared chat-completions path, `reasoning_content` captured |
 | OpenRouter | Yes | Yes | Yes | No | Varies | Meta-provider; `providerOptions.openrouter` for routing prefs |
 | Local | Yes | Yes | Yes | No | No | Any OpenAI-compatible server; endpoint is the catalog entry's `baseURL` |
-| Novita | No | Yes | Yes | No | No | Prices in the model list; text via the shared chat-completions path, separate image adapter |
+| Novita | Yes | Yes | Yes | No | Yes (`reasoning_content`) | Prices in the model list; text via the shared chat-completions path, separate image adapter |
 
 Adapter capability ≠ model capability — whether a given model accepts images, tools, or thinking effort depends on the model spec in `curated.json`. The adapter passes through what the envelope supplies; the provider rejects unsupported combos.
 

package/js/session/adapters/cerebras.js

@@ -17,6 +17,7 @@ import { runChatCompletions } from './_chat_completions.js'
 export async function * cerebras (envelope, deps = {}) {
   const client = deps.client ?? new Cerebras({ apiKey: envelope.auth.key })
   yield * runChatCompletions(envelope, client, {
+    stream: true,
     provider: 'cerebras',
     toolChoiceFlavor: 'cerebras',
     reasoningField: 'cerebras_zai'

package/js/session/adapters/groq.js

@@ -1,5 +1,5 @@
 /**
- * Groq adapter — OpenAI-compatible chat completions, non-streaming.
+ * Groq adapter — OpenAI-compatible chat completions.
  *
  * @module session/adapters/groq
  */
@@ -19,7 +19,10 @@ export async function * groq (envelope, deps = {}) {
     apiKey: envelope.auth.key,
     fetchOptions: { dispatcher: streamingDispatcher() }
   })
-  yield * runChatCompletions(envelope, client, { provider: 'groq' }, {
+  yield * runChatCompletions(envelope, client, {
+    provider: 'groq',
+    stream: true
+  }, {
     signal: deps.signal,
     log: deps.log,
     span: deps.span

package/js/session/adapters/mistral.js

@@ -26,6 +26,7 @@ export async function * mistral (envelope, deps = {}) {
     fetchOptions: { dispatcher: streamingDispatcher() }
   })
   yield * runChatCompletions(envelope, client, {
+    stream: true,
     provider: 'mistral',
     toolChoiceFlavor: 'mistral'
   }, {

package/js/session/adapters/novita.js

@@ -9,6 +9,7 @@
 import OpenAI from 'openai'
 
 import { runChatCompletions } from './_chat_completions.js'
+import { streamingDispatcher } from './_dispatcher.js'
 
 const BASE_URL = 'https://api.novita.ai/openai'
 
@@ -18,9 +19,14 @@ const BASE_URL = 'https://api.novita.ai/openai'
  * @returns {AsyncGenerator<import('#core/events.js').Event>}
  */
 export async function * novita (envelope, deps = {}) {
-  const client = deps.client ?? new OpenAI({ apiKey: envelope.auth.key, baseURL: BASE_URL })
+  const client = deps.client ?? new OpenAI({
+    apiKey: envelope.auth.key,
+    baseURL: BASE_URL,
+    fetchOptions: { dispatcher: streamingDispatcher() }
+  })
   yield * runChatCompletions(envelope, client, {
-    provider: 'novita'
+    provider: 'novita',
+    stream: true
   }, {
     signal: deps.signal,
     log: deps.log,

package/js/session/adapters/qwen.js

@@ -27,6 +27,7 @@ export async function * qwen (envelope, deps = {}) {
     fetchOptions: { dispatcher: streamingDispatcher() }
   })
   yield * runChatCompletions(envelope, client, {
+    stream: true,
     provider: 'qwen',
     reasoningField: 'qwen'
   }, {

package/js/session/adapters/xiaomi.js

@@ -26,6 +26,7 @@ export async function * xiaomi (envelope, deps = {}) {
     fetchOptions: { dispatcher: streamingDispatcher() }
   })
   yield * runChatCompletions(envelope, client, {
+    stream: true,
     provider: 'xiaomi'
   }, {
     signal: deps.signal,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mohdel",
-  "version": "1.0.3",
+  "version": "1.1.0",
   "license": "MIT",
   "author": {
     "name": "Christophe Le Bars",
@@ -135,7 +135,7 @@
     "@opentelemetry/exporter-trace-otlp-grpc": "^0.222.0",
     "@opentelemetry/sdk-node": "^0.222.0",
     "chalk": "^6.0.0",
-    "mohdel-thin-gate-linux-x64-gnu": "1.0.3"
+    "mohdel-thin-gate-linux-x64-gnu": "1.1.0"
   },
   "dependencies": {
     "@anthropic-ai/sdk": "^0.125.0",