Waytrine

Task Router

A 322M-parameter classifier that makes ten routing decisions in one forward pass, served as an API and running in production inside PugPort.

Code: Waytrine/task-router, MIT. Weights: Hugging Face, Apache-2.0.

Code on GitHub Waytrine/task-router ↗

97.8%

Held-out synthetic test, v4. Untuned: 44.0%.

91.0%

Sealed real traffic, v5. v4, in production: 72.0%.

37ms

Median, short request

10

Heads, one forward pass

322M

Parameters

Encoder
mmBERT-base, ModernBERT, 22 layers
License
MIT code, Apache-2.0 weights
Weights
Hugging Face, v4
Training data
Chinese and English
Input
Up to 1024 tokens
Status
v4 in production; v5 evaluated, rolling out next

In production: PugPort

PugPort is the agent side: it runs many AI agents and accounts from one chat. Task Router is the model side: it reads each task and judges its type and difficulty, so PugPort can pick a cheaper model or a stronger one.

Ten decisions, one pass

Task Router reads a bounded state of five parts: the message, recent context, quoted reference text, attachment metadata (kind, name and type, never the contents) and the candidate executors. It answers ten single-choice questions in one forward pass.

PugPort acts on the answers. difficulty picks the model tier and the effort level, task_type picks the routing category, and project_access, run_commands, browser_use and media_output decide whether a message gets a reply or becomes a task.

The ten heads and their answers
HeadAnswers
task_type 10conversation, coding, research, writing, translation, data_analysis, media, automation, mixed, unclear
difficulty 4simple, standard, complex, unclear
web_search 3yes, no, unclear
document_read 3yes, no, unclear
run_commands 3yes, no, unclear
browser_use 3yes, no, unclear
input_modality 6text, image, audio, video, mixed, unclear
media_output 9none, image_create, image_edit, speech, music, video_generate, video_render, mixed, unclear
project_access 4none, read, write, unclear
executor_intent 6none, assign, exclude, reassign, mention_only, unclear

Four rounds, one day

Four fine-tuning rounds on 26 October 2026, each on more data. Held-out accuracy went from 44.0% untuned to 97.8%.

  1. 44.0% Base
    untuned
  2. 79.8% v1
    240 rows
  3. 83.7% v2
    400 rows
  4. 96.2% v3
    520 rows
  5. 97.8% v4
    720 rows

The synthetic dataset has 200 scenarios, each written in Chinese and English: 400 base rows in 169 families, written by an AI assistant and reviewed by agents.

Splits are grouped by family, with zero family overlap across train (240), validation (60), calibration (40) and test (60). A test row never has a sibling in training.

A second check, on an independent 24-item set frozen before inference: v4 scored 93.3%, 224 of 240 decisions.

Held-out synthetic test, 60 rows. Exact: all ten heads right.
ModelTrain rowsMeanExact
Untuned base – 44.0% 0.0%
v1 240 79.8% 21.7%
v2 400 83.7% 31.7%
v3 520 96.2% 70.0%
v4 720 97.8% 78.3%

Soft cross-entropy · 4 epochs · batch 8 × gradient accumulation 2 · learning rate 2.5e-5 encoder, 1e-4 heads · weight decay 0.01 · gradient clip 1.0 · full encoder fine-tune · bf16

97.8% on paper. 72.0% on real traffic.

A high score on synthetic data is a claim about synthetic data. I built a sealed test from real traffic to check it, found the gap, and closed most of it.

  1. First, a test that can't leak

    392 real requests from my own use, in 86 task groups, labelled blind by agents that never saw the model's prediction, with 69 labels cross-checked.

    One item per group went into a sealed holdout of 30, fixed by a SHA-256 manifest. The 132 other records from those groups were quarantined, so nothing similar reached training.

  2. Then, the cause per head

    v4 scored 72.0% over 300 decisions. Seven heads scored 70% or more; three did not: difficulty 33.3%, project_access 26.7%, run_commands 40.0%.

    The synthetic data under-represented long coding work that needs write access and commands, so the model rated most real work “standard”. On 154 historical production judgments, its difficulty grade agreed with reviewers 24.0% of the time.

  3. Finally, 75 real rows

    v5 retrained with 75 reviewed real rows, with 35 more held for development. Three variants scored 89.3%, 90.3% and 91.0% on the same sealed 30.

    The best lifted difficulty to 80.0%, project_access and run_commands to 96.7%. The synthetic regression sets did not drop: 98.0%, 99.0%, 99.5%.

v4 72.0%

v5, trained from the base model 91.0%

project_access 26.7 96.7
difficulty 33.3 80.0
run_commands 40.0 96.7
document_read 70.0 73.3
executor_intent 73.3 80.0
task_type 86.7 93.3
browser_use 96.7 96.7
input_modality 96.7 96.7
web_search 96.7 96.7
media_output 100.0 100.0
Accuracy per head on the sealed holdout: 30 real requests, 300 decisions. Rows with all ten heads right: 1 of 30 for v4, 12 of 30 for the selected v5.

Within 3 points of a model 12× its size

Same schema, same sealed 30 items. Task Router v5 lands within 3 points of models 2.5× and 12× its size.

Sealed real-traffic holdout, mean accuracy over the ten heads
ModelSizeSetupAccuracy
H2O-Lightning-4B 4B LoRA 94.0%
Intern-Decision-0.8B 0.8B Fine-tuned 93.0%
Task Router v5 322M Fine-tuned 91.0%
elephant-1 9B Zero-shot 90.3%
GLiNER2.5-multi 287M Fine-tuned 83.0%

The API: bounded, pinned, forgetful

POST /v1/judge and GET /healthz. The app never waits past its deadline, and the server never queues more than eight jobs.

  1. Client

    A deadline, then rule-based routing. A 30 s back-off after an error. Pins the model version and the weight SHA-256.

  2. Edge

    Cloudflare Tunnel to a server bound to loopback. Per-IP rate limit.

  3. Server

    A raw ASGI app under Uvicorn, one worker.

  4. Queue

    One inference at a time, 8 waiting. Full: 503 with Retry-After. Waited over 2.5 s: dropped.

  5. Model

    Apple silicon, MPS, FP32. A warm-up prediction after 60 s idle. Every response carries the version and weight hash.

37ms

Median, short request (Apple silicon, MPS)

68ms

Median, medium request (Apple silicon, MPS)

140ms

Median, long request (Apple silicon, MPS)

82ms

Warm median, RTX 4060 laptop, p95 107 ms

Read the API reference Endpoints, request and response schemas, every status code, limits, curl and TypeScript examples →

Nothing kept that isn't needed

Logs keep time, status, latency, token count and country. Request bodies are not kept. Before anything leaves the app, credentials are redacted by pattern (private keys, auth headers, common provider API keys, JWTs, bearer tokens, URL credentials) and file paths are removed.

Sharing samples for training is optional: on by default, off by default in the EU/EEA, the UK and Switzerland, under a random install ID with a delete token whose hash is all the server stores.

Advice, never authority

The model's advice can never widen permissions. Explicit user pins, approved rules, quotas, budgets and read-only limits always win. When the advice conflicts with them, the conflict is recorded and the advice dropped.

Where it runs

It routes requests in PugPort, a multi-agent desktop app. Code: Waytrine/task-router (MIT), under the Waytrine organization. The v4 weights are on Hugging Face (Apache-2.0).