Merge pull request #4477 from janhq/fix/remote-engine-request-templates

chore: fix remote engines request templates

Author: louis-jan

extensions/engine-management-extension/models/martian.json

@@ -5,7 +5,7 @@
     "name": "Martian Model Router",
     "version": "1.0",
     "description": "Martian Model Router dynamically routes requests to the best LLM in real-time",
-    "parameters": {
+    "inference_params": {
       "max_tokens": 4096,
       "temperature": 0.7,
       "top_p": 0.95,

extensions/engine-management-extension/models/mistral.json

@@ -5,7 +5,7 @@
     "name": "Mistral Small",
     "version": "1.1",
     "description": "Mistral Small is the ideal choice for simple tasks (Classification, Customer Support, or Text Generation) at an affordable price.",
-    "parameters": {
+    "inference_params": {
       "max_tokens": 32000,
       "temperature": 0.7,
       "top_p": 0.95,
@@ -19,7 +19,7 @@
     "name": "Mistral Large",
     "version": "1.1",
     "description": "Mistral Large is ideal for complex tasks (Synthetic Text Generation, Code Generation, RAG, or Agents).",
-    "parameters": {
+    "inference_params": {
       "max_tokens": 32000,
       "temperature": 0.7,
       "top_p": 0.95,
@@ -33,7 +33,7 @@
     "name": "Mixtral 8x22B",
     "version": "1.1",
     "description": "Mixtral 8x22B is a high-performance, cost-effective model designed for complex tasks.",
-    "parameters": {
+    "inference_params": {
       "max_tokens": 32000,
       "temperature": 0.7,
       "top_p": 0.95,

extensions/engine-management-extension/models/nvidia.json

@@ -5,7 +5,7 @@
     "name": "Mistral 7B",
     "version": "1.1",
     "description": "Mistral 7B with NVIDIA",
-    "parameters": {
+    "inference_params": {
       "max_tokens": 1024,
       "temperature": 0.3,
       "top_p": 1,

extensions/engine-management-extension/models/openrouter.json

@@ -5,7 +5,7 @@
     "name": "OpenRouter",
     "version": "1.0",
     "description": " OpenRouter scouts for the lowest prices and best latencies/throughputs across dozens of providers, and lets you choose how to prioritize them.",
-    "parameters": {
+    "inference_params": {
       "max_tokens": 128000,
       "temperature": 0.7,
       "top_p": 0.95,

extensions/engine-management-extension/resources/anthropic.json

@@ -15,7 +15,7 @@
     },
     "transform_resp": {
       "chat_completions": {
-        "template": "{% if input_request.stream %} {\"object\": \"chat.completion.chunk\", \"model\": \"{{ input_request.model }}\", \"choices\": [{\"index\": 0, \"delta\": { {% if input_request.type == \"message_start\" %} \"role\": \"assistant\", \"content\": null {% else if input_request.type == \"ping\" %} \"role\": \"assistant\", \"content\": null {% else if input_request.type == \"content_block_delta\" %} \"role\": \"assistant\", \"content\": \"{{ input_request.delta.text }}\" {% else if input_request.type == \"content_block_stop\" %} \"role\": \"assistant\", \"content\": null {% else if input_request.type == \"content_block_stop\" %} \"role\": \"assistant\", \"content\": null {% endif %} }, {% if input_request.type == \"content_block_stop\" %} \"finish_reason\": \"stop\" {% else %} \"finish_reason\": null {% endif %} }]} {% else %} {\"id\": \"{{ input_request.id }}\", \"created\": null, \"object\": \"chat.completion\", \"model\": \"{{ input_request.model }}\", \"choices\": [{ \"index\": 0, \"message\": { \"role\": \"{{ input_request.role }}\", \"content\": \"{% if input_request.content and input_request.content.0.type == \"text\" %} \"{{input_request.content.0.text}}\" {% endif %}\", \"refusal\": null }, \"logprobs\": null, \"finish_reason\": \"{{ input_request.stop_reason }}\" } ], \"usage\": { \"prompt_tokens\": {{ input_request.usage.input_tokens }}, \"completion_tokens\": {{ input_request.usage.output_tokens }}, \"total_tokens\": {{ input_request.usage.input_tokens + input_request.usage.output_tokens }}, \"prompt_tokens_details\": { \"cached_tokens\": 0 }, \"completion_tokens_details\": { \"reasoning_tokens\": 0, \"accepted_prediction_tokens\": 0, \"rejected_prediction_tokens\": 0 } }, \"system_fingerprint\": \"fp_6b68a8204b\"} {% endif %}"
+        "template": "{% if input_request.stream %} {\"object\": \"chat.completion.chunk\", \"model\": \"{{ input_request.model }}\", \"choices\": [{\"index\": 0, \"delta\": { {% if input_request.type == \"message_start\" %} \"role\": \"assistant\", \"content\": null {% else if input_request.type == \"ping\" %} \"role\": \"assistant\", \"content\": null {% else if input_request.type == \"content_block_delta\" %} \"role\": \"assistant\", \"content\": \"{{ input_request.delta.text }}\" {% else if input_request.type == \"content_block_stop\" %} \"role\": \"assistant\", \"content\": null {% else if input_request.type == \"content_block_stop\" %} \"role\": \"assistant\", \"content\": null {% endif %} }, {% if input_request.type == \"content_block_stop\" %} \"finish_reason\": \"stop\" {% else %} \"finish_reason\": null {% endif %} }]} {% else %} {\"id\": \"{{ input_request.id }}\", \"created\": null, \"object\": \"chat.completion\", \"model\": \"{{ input_request.model }}\", \"choices\": [{ \"index\": 0, \"message\": { \"role\": \"{{ input_request.role }}\", \"content\": {% if input_request.content and input_request.content.0.type == \"text\" %} \"{{input_request.content.0.text}}\" {% else %} null {% endif %}, \"refusal\": null }, \"logprobs\": null, \"finish_reason\": \"{{ input_request.stop_reason }}\" } ], \"usage\": { \"prompt_tokens\": {{ input_request.usage.input_tokens }}, \"completion_tokens\": {{ input_request.usage.output_tokens }}, \"total_tokens\": {{ input_request.usage.input_tokens + input_request.usage.output_tokens }}, \"prompt_tokens_details\": { \"cached_tokens\": 0 }, \"completion_tokens_details\": { \"reasoning_tokens\": 0, \"accepted_prediction_tokens\": 0, \"rejected_prediction_tokens\": 0 } }, \"system_fingerprint\": \"fp_6b68a8204b\"} {% endif %}"
       }
     }
   }

extensions/engine-management-extension/resources/mistral.json

@@ -10,12 +10,12 @@
     "transform_req": {
       "chat_completions": {
         "url": "https://api.mistral.ai/v1/chat/completions",
-        "template": "{ {% set first = true %} {% for key, value in input_request %} {% if key == \"messages\" or key == \"model\" or key == \"temperature\" or key == \"store\" or key == \"max_tokens\" or key == \"stream\" or key == \"presence_penalty\" or key == \"metadata\" or key == \"frequency_penalty\" or key == \"tools\" or key == \"tool_choice\" or key == \"logprobs\" or key == \"top_logprobs\" or key == \"logit_bias\" or key == \"n\" or key == \"modalities\" or key == \"prediction\" or key == \"response_format\" or key == \"service_tier\" or key == \"seed\" or key == \"stop\" or key == \"stream_options\" or key == \"top_p\" or key == \"parallel_tool_calls\" or key == \"user\" %} {% if not first %},{% endif %} \"{{ key }}\": {{ tojson(value) }} {% set first = false %} {% endif %} {% endfor %} }"
+        "template": "{{tojson(input_request)}}"
       }
     },
     "transform_resp": {
       "chat_completions": {
-        "template": "{ {% set first = true %} {% for key, value in input_request %} {% if key == \"choices\" or key == \"created\" or key == \"model\" or key == \"service_tier\" or key == \"system_fingerprint\" or key == \"stream\" or key == \"object\" or key == \"usage\" %} {% if not first %},{% endif %} \"{{ key }}\": {{ tojson(value) }} {% set first = false %} {% endif %} {% endfor %} }"
+        "template": "{{tojson(input_request)}}"
       }
     }
   }

web/utils/messageRequestBuilder.ts

@@ -167,7 +167,6 @@ export class MessageRequestBuilder {
       messages: this.normalizeMessages(this.messages),
       model: this.model,
       thread: this.thread,
-      engine: this.model.engine,
     }
   }
 }