Understand this story
On November 6, 2026, OpenRouter released a new generation of Auto Router and added Auto Code (openrouter/auto-code), designed specifically for coding scenarios; the old slug openrouter/auto remains unchanged. According to the official announcement, the new router was rebuilt on an entirely new foundation and improves by up to 4x over the old version on some benchmarks, with p90 routing-step latency reduced from 1.96s to 0.65s; Auto Code matches Claude Opus 5.5 on five benchmarks at 74% lower cost per task. The new router synthesizes 11 request features and combines real-time pricing, provider health, and cache status to select models; users can set model to openrouter/auto or openrouter/auto-code to call it directly. On November 7, OpenRouter further explained that Auto Code is aimed at coding agents and can display the routing decision for each request; the company says that during testing, invalid tool calls were reduced by 4x, agent response truncation decreased by 50%, and latency was reduced by 3x compared with the previous-generation router.
AI-generated from public reports
Latest development · Report
OpenRouter Launches Next-Generation Auto Router and Auto Code Model RoutersDevelopments
1 confirmed developments
OpenRouter launches new-generation Auto Router and Auto Code
Source:OpenRouter：Announcements
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 7
OpenRouterFirst-party
OpenRouter Launches Next-Generation Auto Router and Auto Code Model Routers
OpenRouter has released a new generation of model routers, including Auto Router for general Q&A and agents, and Auto Code for coding agents. The company claims that Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of only 0.65s, and it can show the basis for routing decisions on a per-request basis. Internal testing data shows that, compared with the previous generation of routers, invalid tool calls are reduced by 4x, truncated agent responses are reduced by 50%, and latency is reduced by 3x; Auto Router processes over 2.5T tokens per month from more than 100,000 people.
OpenRouterFirst-party
OpenRouter Launches Next-Generation Model Routers Auto Router and Auto Code
OpenRouter has released a new generation of model routers, including Auto Router for general Q&A and agents, and Auto Code for coding agents. The company claims that Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of only 0.65s; during testing, compared with the previous-generation router, invalid tool calls were reduced by 4x, agent response truncation decreased by 50%, and latency was reduced by 3x. The new routers support showing the basis for routing decisions on a per-request basis, and will still seek the most cost-optimal option when the model list is restricted.
OpenRouterFirst-party
OpenRouter Launches Auto Router and Auto Code, a New Generation of Model Routers
OpenRouter has released a new generation of model routers, Auto Router and Auto Code, aimed respectively at general Q&A and Agents, and at coding Agents. The company says Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of only 0.65s, and it can show the routing decision process per request. During testing, compared with the previous generation of routers, invalid tool calls dropped 4x, Agent response truncation fell by 50%, and latency was reduced 3x; Auto Router processes 2.5T tokens per month from more than 100,000 people.
OpenRouterFirst-party
OpenRouter Launches Auto Router and Auto Code, a New Generation of Model Routers
OpenRouter has released a new generation of model routers: Auto Router (for general Q&A and Agents) and Auto Code (for coding Agents). According to the official announcement, Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of only 0.65s, and it can display routing decisions per request. Once a new model joins OpenRouter, it immediately enters the candidate pool; the router automatically competes based on benchmarks and popularity signals to select the most cost-effective combination.
OpenRouterFirst-party
OpenRouter Launches Next-Generation Auto Router and Auto Code Model Routers
OpenRouter has released a new generation of model routers, Auto Router and Auto Code; the former targets general-purpose Q&A and agents, while the latter targets coding agents. The company says Auto Code's benchmarks are close to Opus 5.5, at 74% lower cost, with p90 latency of just 0.65s, and it can display routing decisions per request. During the testing period, compared with the previous-generation router, invalid tool calls dropped 4x, agent response truncation fell by 50%, and latency improved 3x; Auto Router processes over 2.5T tokens per month from more than 100,000 users.
OpenRouterFirst-party
OpenRouter Launches Next-Generation Auto Router and Auto Code Model Routers
OpenRouter has released a new generation of model routers, including Auto Router for general Q&A and agents, and Auto Code for coding agents. The company says Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of just 0.65s; during the testing period, invalid tool calls dropped 4x, agent response truncation fell by 50%, and latency dropped to one-third that of the previous-generation router. Users can enable router metadata in the chat room and API to see which models were considered for each request, why one was selected, and what factors influenced the decision.
OpenRouterFirst-party
OpenRouter Launches Next-Generation Auto Router and Auto Code Model Routers
OpenRouter has released a new generation of model routers, including Auto Router for general-purpose Q&A and Agents, and Auto Code for coding Agents. The company claims that Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of only 0.65s; when customers' available models are restricted by policies such as region and data retention, Auto Router can still select the most cost-effective option within the restricted model pool. During the testing period, compared with the previous-generation router, invalid tool calls were reduced by 4x, Agent response truncation decreased by 50%, and latency was reduced by 3x, and the reasons for model selection can be viewed via router metadata in the chat room and API.
OpenRouterFirst-party
OpenRouter Launches Next-Generation Auto Router and Auto Code Model Routers
OpenRouter has released a new generation of model routers: Auto Router (for general Q&A and Agents) and Auto Code (for coding Agents). The company says Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of just 0.65s, and it can show routing decisions per request. During testing, compared with the previous-generation router, invalid tool calls dropped 4x, Agent response truncation fell by 50%, and latency improved 3x; Auto Router processes 2.5T tokens per month generated by over 100k users, and all benchmark results are published alongside the best single-model baseline and other routers.
OpenRouterFirst-party
OpenRouter Launches Auto Router and Auto Code, a New Generation of Model Routers
OpenRouter has released a new generation of model routers: Auto Router (for general Q&A and agents) and Auto Code (for coding agents). According to the company, Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of 0.65s, and it can show the basis for routing decisions on a per-request basis. During testing, compared with the previous generation of routers, invalid tool calls were reduced 4x, agent response truncation decreased by 50%, and latency was reduced 3x; Auto Router processes over 2.5T tokens per month from more than 100,000 users.
OpenRouterFirst-party
OpenRouter Releases a New Generation of Model Routers: Auto Router and Auto Code
OpenRouter has launched a new generation of model routers: Auto Router (for general Q&A and agents) and Auto Code (for coding agents). According to the official announcement, Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of only 0.65s, and it can show the basis for routing decisions on a per-request basis. Auto Router processes 2.5T tokens generated by over 100k users each month; during the testing period, invalid tool calls were reduced by 4x, agent response truncation decreased by 50%, and latency was reduced by 3x compared with the previous generation of routers.
OpenRouterFirst-party
OpenRouter Launches Next-Generation Model Routers Auto Router and Auto Code
OpenRouter has released a new generation of model routers, including Auto Router for general Q&A and agents, and Auto Code for coding agents. Auto Code's benchmark performance is close to Opus 5.5, at 74% lower cost, with p90 latency of 0.65s, and it can display the routing decision for each request. The company says that during testing, invalid tool calls were reduced by 4x, agent response truncation decreased by 50%, and latency was reduced by 3x compared with the previous generation of routers.
Nov 6
OpenRouter：AnnouncementsFirst-partySelected
OpenRouter Launches a New Generation of Auto and Auto Code Routers
OpenRouter has released a new generation of Auto Router and added Auto Code (openrouter/auto-code), designed specifically for coding scenarios; the legacy slug openrouter/auto remains unchanged. According to the official announcement, the new router is rebuilt on an entirely new foundation, delivering up to a 4x improvement over the previous version on some benchmarks, with p90 latency for the routing step reduced from 1.96s to 0.65s; Auto Code matches Claude Opus 5.5 on five benchmarks while costing 74% less per task. The new router selects models by synthesizing 11 request features combined with real-time pricing, provider health, and cache status; users can invoke it directly by setting model to openrouter/auto or openrouter/auto-code.