iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
LiteLLM Auto Router lets a client call one model name while the gateway selects a configured model for each prompt. You can set it up in the dashboard or in config.yaml; the key is mapping routing tiers to model names your deployment already serves. The feature is documented as an add-on, so first confirm it is available in your LiteLLM version and environment.
What Auto Router does—and what it does not
Auto Router classifies a prompt into a tier, then sends it to the model mapped to that tier. Your application can keep calling a single router name, such as smart-router, instead of choosing a destination model for every request.
This is different from deployment load balancing. Auto Router chooses which model is appropriate for a prompt; LiteLLM’s router and load-balancing features distribute requests among deployments and handle concerns such as retries, cooldowns, and fallbacks. See LiteLLM’s Router – Load Balancing documentation for those separate behaviors.
Free tools Windows power users keep installed
One-click scans. No signup required.
Availability may differ by LiteLLM release or account because the official Auto Router documentation describes it as an add-on and invites early-access/design-partner participation. Check your own dashboard and installation before planning a rollout.
#1 Best Overall
- EVOLUTION AMD RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
Before configuring the router
- Confirm that each destination model is already configured and served by your LiteLLM gateway.
- Choose the actual model names from your configuration; tier mappings must refer to those names, not necessarily provider model identifiers.
- Decide which configured model should receive requests when you set the router’s default.
- Identify representative prompts for each kind of request your application handles, including straightforward tasks and demanding reasoning work.
Set up Auto Router in the dashboard
- Open Models + Endpoints in the LiteLLM dashboard and choose to add a model.
- Select Auto Router.
- Choose the automatic configuration flow or start from a template.
- Review the routing tiers and the destination model assigned to each. Confirm that every destination name matches a model available in your gateway.
- Use Test Routing with representative prompts and inspect the selected destinations.
- Save the configuration after reviewing the results.
Dashboard labels and availability can vary with the LiteLLM release and environment. If Auto Router is not offered in your dashboard, check the official documentation for the setup options supported by your installation.
Configure Auto Router in config.yaml
The documented YAML integration uses auto_router/complexity_router. In this example, replace the illustrative provider/model identifiers and API-key environment variable with values supported by your environment. The names small-model and stronger-model are LiteLLM model names defined in the same configuration, and the tier values refer to those names.
Rank #2
- Built for Local AI Development: AMD Ryzen AI Halo is designed for local AI development and inference, featuring 128GB unified memory and support for up to 200B parameter models to build and run intensive AI workloads locally.
- 128GB Unified Memory: Features 128GB LPDDR5x unified memory at 8000 MT/s with 256 GB/s memory bandwidth, providing a shared memory pool across the CPU, GPU, and NPU to support larger AI models.
- AMD Ryzen AI Max+ 395 Processor: Features 16 cores, 32 threads, and Zen 5 architecture, paired with AMD Radeon 8060S integrated graphics featuring 40 RDNA 3.5 compute units and an AMD XDNA 2 NPU with up to 50 TOPS.
- Linux AI Developer Platform: Purpose-built for Linux-based AI development with full AMD ROCm software support and preloaded tools, models, and workflows optimized for local AI development.
- Compact, Connected Design: Includes a 2TB M.2 SSD, 10GbE LAN, Wi-Fi 7, Bluetooth 5.4, USB-C connectivity, and HDMI 2.1b.
model_list:
- model_name: small-model
litellm_params:
model: provider/small-model
api_key: os.environ/PROVIDER_API_KEY
- model_name: stronger-model
litellm_params:
model: provider/stronger-model
api_key: os.environ/PROVIDER_API_KEY
- model_name: smart-router
litellm_params:
model: auto_router/complexity_router
complexity_router_config:
tiers:
SIMPLE: small-model
MEDIUM: stronger-model
COMPLEX: stronger-model
REASONING: stronger-model
classifier_type: heuristic
complexity_router_default_model: stronger-modelHere, clients request smart-router. LiteLLM uses the configured classifier to select a tier, looks up that tier’s destination, and uses stronger-model as the default specified by complexity_router_default_model. Choose a default that exists in your configuration and is suitable for requests that do not resolve to another destination.
Recommended Free Tools
The tier names and mappings above illustrate the documented pattern, not a required allocation. Adjust them to fit the models your deployment serves and the tasks you want each to handle.
Rank #3
- EVOLUTION RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
Choose a classifier for your routing policy
LiteLLM documents several classifier approaches: heuristics, an LLM, JEV through TypeSafe System One Choice, keyword rules, and custom plugins. The available evidence does not establish one as universally most accurate, fastest, or least expensive.
- Heuristics: A documented option used in the YAML example. Validate its tier choices against your own prompt mix.
- LLM: Use an LLM-based classifier if it fits your architecture; measure its behavior and operational impact with your traffic.
- JEV: LiteLLM lists JEV through TypeSafe System One Choice as an option. Confirm its integration requirements in the current documentation.
- Keyword rules: Consider these when explicit terms or patterns are useful signals, and test edge cases where wording may not reflect task difficulty.
- Custom plugins: Use a custom approach when your routing policy needs signals or logic not covered by the built-in choices.
Compare options on the dimensions that matter to your service—routing quality, latency, and cost—rather than assuming a classifier will produce a particular result.
Test routing before sending production traffic
A saved configuration is not evidence that the tiers fit your workload. Use the dashboard’s routing test, then follow LiteLLM’s recommendation to shadow-evaluate against your own traffic before switching production requests. Compare which models are selected and assess the resulting answer quality, latency, and cost. These outcomes depend on your prompts, models, and configuration; the documentation does not establish a guaranteed accuracy or savings figure.
After evaluation, use savings accounting to understand the financial effect of the routing policy. If results are not acceptable, revise the tier mappings, classifier, or default model and evaluate again before making the router the production path.
Quick Recap
Common setup mistakes to avoid
- Mapping a tier to a provider identifier rather than a configured LiteLLM name: Use the model names that appear in your
model_list. - Leaving the default to chance: Set
complexity_router_default_modelto a destination that is present and intentionally chosen for default requests. - Expecting Auto Router to manage deployment distribution: Use LiteLLM’s separate load-balancing configuration for deployment-level distribution, retries, cooldowns, and fallbacks.
- Assuming the example’s provider models are universally available: The identifiers are illustrative; substitute models supported by your provider and environment.
- Switching traffic based on a few convenient prompts: Test a representative sample and shadow-evaluate your own traffic before production use.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
