Neurometric Task Router
Intelligent Routing
On All Your Tasks
Other routers guess from price tags and leaderboards. Task Router learns from your workloads so your inference is always optimized for your requirements.
Keep your OpenAI SDK. Change the base URL and model alias.
See routing in action
Your requirements choose the model. #
Choose a task and set your quality target. See which models qualify, which one is selected, and the fallback order—all from published benchmark evidence.
- 1 · Your task Quick-Reply
- 2 · Your requirements 90% minimum quality · lowest cost
- 3 · Model selection Qwen3 4B Instruct 2507
- 4 · Stable API endpoint taskrouter/<your-route>
Lowest-cost policy. Quality is measured on the published evaluation set; it is not a guarantee for every request.
Sign in to save your route. Your task and requirements come with you.
Selected model
Qwen3 4B Instruct 2507
Selected because it meets your 90% quality target at the lowest measured cost.
Fallback order: Gemma 4 E4B IT → Ministral 8B Instruct 2410 → Granite 4.1 8B
| Measured inference cost for 100,000 ticket. Routing fees are additional. | ||||
|---|---|---|---|---|
| Model | Quality | p50 | Cost | Selection |
| --- | --- | --- | --- | --- |
| Nemotron Nano 9B v2 | 100.0% | 2.33s | $9.402 | Qualifies |
| Qwen3 235B A22B | 99.0% | 1.12s | $11.198 | Qualifies |
| DeepSeek V3 | 100.0% | 1.07s | $24.696 | Qualifies |
| Mistral Large 2407 | 100.0% | 1.47s | $153.30 | Qualifies |
| Arcee Trinity Large Thinking | 99.0% | 1.41s | $42.08 | Qualifies |
| GPT-5.4 | 100.0% | 1.31s | $145.50 | Qualifies |
| Claude Opus 4.7 | 100.0% | 2.04s | $490.60 | Qualifies |
| Gemini 3.1 Pro Preview | 100.0% | 3.97s | $455.60 | Qualifies |
| Gemma 4 E4B IT | 95.0% | 0.47s | $1.0234 | Fallback 1 |
| Granite 4.1 8B | 99.0% | 0.72s | $2.13 | Fallback 3 |
| Ministral 8B Instruct 2410 | 100.0% | 1.39s | $1.3061 | Fallback 2 |
| Qwen3 4B Instruct 2507 | 100.0% | 1.80s | $0.4906 | Selected |
Evidence preview, not a live request. Route creation checks current availability; your API alias stays stable as qualified models change.
Why task-aware routing
Model choice is a task question, not a model question. #
A model can excel at one workload and fall short on another. General leaderboards cannot tell you which. Task-level measurements can.
Choosing a model yourself
- Compare general benchmarks and price tags
- Build and run your own task evaluations
- Review each new model and update your integration
- Maintain your own fallback chain
With Task Router
- Inspect published evidence for the task you are running
- Set a quality floor and an optional latency cap
- Keep a stable route alias as qualified models change
- Use an ordered fallback chain of qualified models
For developers
Two changes to your code. #
Evidence already built.
- 1### Create a route for your taskPick a published task, set your quality floor, and choose a routing preset.
- 2### Point your OpenAI SDK at itUse your organization API key and the stable taskrouter/<slug> alias.
- 3### Ship, then track the routeWith auto-optimize on, the route can adapt as qualified models and prices change.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.neurometric.ai/v1",
api_key=os.environ["NEUROMETRIC_API_KEY"],
)
response = client.chat.completions.create(
model="taskrouter/privilege-triage",
messages=[{"role": "user", "content": email_thread}],
)
Replace taskrouter/privilege-triage with the alias shown on your route.
Pricing
Free #
for 1,000 routing decisions per month
Then $0.15 per 1,000 decisions
Plus inference at the provider's price, with no markup. Routing only pays for itself if it saves you money, so here's the math on one measured task.
100K emails
$666.00
$15.21
$14.85
$635.94 · 95%
Not ready to route yet? #
Task Explorer is free. Describe a task, compare measured quality, cost, and speed, and inspect the evidence before you commit.
Questions developers ask #
What if my task is not supported? #
Describe it in Task Explorer to find related published evidence. If nothing fits, submit a benchmark request for review.
Is it OpenAI-compatible? #
Use the OpenAI SDK, set the base URL to api.neurometric.ai/v1, and use the model alias shown on your route.
What if no model clears my quality floor? #
You cannot activate a new route without a qualified model. The optimizer keeps an existing route’s last-good plan when it cannot qualify a replacement.
How does this work with OpenRouter? #
OpenRouter is a serving provider. Task Router adds task-specific benchmark evidence and a policy that orders qualified model candidates.
What can I control? #
Choose lowest cost, balanced, or highest quality, set a quality floor and p50 latency cap, and turn automatic optimization on or off.
Do I need to sign in to explore? #
No. Explorer, Tasks, Models, and Providers are public. Sign in to run Playground requests, save a watchlist, or create production routes.