Jev Router, measured: which model it picks and what it costs
Jev Router is TypeSafe AI's model router on OpenRouter: one model id that chooses a chat model and a reasoning effort for every request. We sent it 24 requests on 2026-10-02, from a one-word fact to a proof, and recorded every choice it made.
Where can I try TypeSafe AI's Jev Router and see which model it selects?
Use typesafe/jev-router on OpenRouter with an OpenRouter account and credits. Jagent's page shows a dated 24-request test with the selected models, response times and billed costs. To choose from your own model tiers instead, use the separate Jev model-routing example.
What does Jev Router do, and what did it cost?
Jev Router uses Jev, TypeSafe's decision model, to read each request and choose which chat model should answer it and how hard that model should think. You send an ordinary chat request to one model id; you get back an ordinary chat answer, from whichever model Jev Router picked.
| Model id | typesafe/jev-router — listed on OpenRouter 2026-09-25 |
|---|---|
| Price | the chosen model's list price — no routing fee in our run |
| Our run | 24 requests, $0.0056 in total |
| Median time | 1.27 s — end to end, including the chosen model |
| Objective answers | 15 of 16 right — the miss was cut off by our token cap |
Quick answer · verified 2026-10-02
What Jev Router is
OpenRouter's listing describes Jev Router as picking the best model and reasoning effort for each request, balancing quality, speed and cost, and adapting as a conversation evolves. It runs on Jev, the System One model TypeSafe launched on September 15, 2026. Jev itself never writes text; in Jev Router it makes one decision — which model — and the model it chose writes the reply.
It accepts the usual chat parameters, including tools, structured outputs and a reasoning setting, and OpenRouter lists text, image, file, audio and video as inputs; our run sent text only. TypeSafe has not published which models are in its pool or how it scores them, which is why the rest of this page is a measurement rather than a summary.
How to call Jev Router
Jev Router is reached like any OpenRouter model: the OpenAI-compatible endpoint with your OpenRouter key and the model id. Ask for usage to see what each call cost; the response's model field names the model that actually answered.
curl https://openrouter.ai/api/v1/chat/completions \
-H "Authorization: Bearer $OPENROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "typesafe/jev-router",
"messages": [{"role": "user", "content": "Prove that the square root of 2 is irrational."}],
"max_tokens": 2000,
"usage": {"include": true}
}'Leave max_tokens generous. Jev Router can choose a reasoning model, and reasoning tokens count against the same limit — the one miss in our run below is exactly that.
What Jev Router picked, request by request

| Request | Jev Router chose | Time | Billed | Answer |
|---|---|---|---|---|
| Capital of Australia | GPT-6 Luna | 2.3 s | $0.00000 | right |
| Planet closest to the Sun | GPT-6 Luna | 1.0 s | $0.00000 | right |
| Good morning in French | DeepSeek V4.1 Flash | 0.7 s | $0.00007 | right |
| 17 × 23 | GPT-6 Luna | 1.4 s | $0.00001 | right |
| 1234 + 5678 | Gemini 3.8 Flash | 2.7 s | $0.00025 | right |
| Pens at 3 for $2 | GPT-6 Luna | 1.1 s | $0.00002 | right |
| A three-term syllogism | GPT-6 Luna | 1.1 s | $0.00002 | right |
| Review: positive or negative | GPT-6 Luna | 1.3 s | $0.00001 | right |
| Pull an email from a sentence | GPT-6 Luna | 1.4 s | $0.00001 | right |
| English into Japanese | DeepSeek V4.1 Flash | 0.7 s | $0.00019 | open-ended |
| One-sentence summary | GPT-6 Luna | 1.1 s | $0.00002 | open-ended |
| List into a markdown table | DeepSeek V4.1 Flash | 0.8 s | $0.00009 | right |
| Fibonacci in Python | GPT-6 Luna | 4.4 s | $0.00003 | right |
| Fix a one-line bug | GPT-6 Luna | 1.4 s | $0.00002 | right |
| Top five customers in SQL | DeepSeek V4.1 Flash | 1.0 s | $0.00030 | right |
| Haiku about rain | GPT-6 Luna | 1.3 s | $0.00001 | open-ended |
| Names for a coffee shop | GPT-6 Luna | 1.0 s | $0.00002 | open-ended |
| Questions about a job offer | GPT-6 Luna | 1.1 s | $0.00006 | open-ended |
| Inflation, asked in Chinese | GPT-6 Luna | 1.0 s | $0.00002 | open-ended |
| Proof that √2 is irrational | Claude Sonnet 5.5 | 1.9 s | $0.00239 | right |
| Divisors of 360 | GPT-6 Luna | 1.3 s | $0.00002 | right |
| Sum of multiples of 3 or 5 | Gemini 3.8 Flash | 2.9 s | $0.00151 | cut off at our cap |
| Postgres migration plan | DeepSeek V4.1 Flash | 0.8 s | $0.00050 | cut off at our cap |
| Kyoto trip, third turn | GPT-6 Luna | 1.3 s | $0.00007 | open-ended |
The pattern in what Jev Router picked is plain. For short facts, arithmetic, classification, code and even the question in Chinese, Jev Router picked GPT-6 Luna, the cheapest model it used. The French greeting, the translation, the SQL, the table and the migration plan went to DeepSeek V4.1 Flash. Jev Router picked Gemini 3.8 Flash twice, on arithmetic and counting, and once, for the proof, Claude Sonnet 5.5 — the most expensive call in the run, at about a quarter of a cent.
What Jev Router cost
For all 24 requests together Jev Router billed $0.0056, and every call matched the list price of the model that answered it, so the router added no fee of its own. Priced on a single model, the same token counts would have cost $0.0468 on Claude Opus 5.5 and $0.0234 on GPT-6.1 Sol — 8.3 and 4.1 times as much. On GPT-6 Luna alone they would have cost $0.0012, a fifth of what Jev Router billed.
That last comparison is the honest one. Jev Router saves money against sending everything to a flagship, not against sending everything to the cheapest model; what it buys you is the escalation — the proof went to a stronger model without anyone writing a rule for it. Other models would use other token counts, so read those figures as a guide. To price your own traffic, use the LLM cost calculator.
Where Jev Router went wrong
Once. For "Find the sum of all integers n with 1 ≤ n ≤ 100 that are divisible by 3 or 5. Give the number." Jev Router picked Gemini 3.8 Flash, which spent 385 of its 396 output tokens reasoning and reached our 400-token cap before writing the answer. That is our limit, not a wrong answer — but it is the failure you will see in production if you cap replies tightly, because you do not choose the model and some of the models Jev Router picks think before they answer. Every other request with a checkable answer came back right: 15 of 16.
Jev Router or Jev itself?
Use Jev Router when you want an answer and would rather not maintain routing rules. Call Jev directly when you want the decision: your own list of models or tools as options, your own confidence threshold, and the choice logged in your code, at $0.042 per million input tokens for the decision alone (worked through on Jev pricing). The model routing use case shows that pattern with measured output, and Jev in an agent applies it to tool calls.
Skip both when every request needs the same model for consistency or compliance, or when your traffic is all easy — in our run, one cheap model would have been cheaper still.
Jev Router questions
Does Jev Router really reduce LLM API costs?
It depends on your baseline. Our 24-request run on October 2, 2026 cost $0.0056: the same recorded token counts priced on Claude Opus 5.5 would have cost $0.0468, while GPT-6 Luna alone would have cost $0.0012. Those are same-token price estimates, not identical-output experiments. The run shows escalation to stronger models, not a guarantee of savings over every fixed model.
Method and evidence · Checked 2026-10-02
Does Jev Router add a routing fee on OpenRouter?
OpenRouter says requests are billed at the price of the model that serves them. Our October 2, 2026 run matched those model list prices across all 24 calls, with no extra routing fee in the recorded charges. A new run can select different models and therefore have a different total cost.
Source · Checked 2026-10-04
Which models does Jev Router choose for different tasks?
In our October 2, 2026 text-only run it selected GPT-6 Luna for 16 requests, DeepSeek V4.1 Flash for 5, Gemini 3.8 Flash for 2 and Claude Sonnet 5.5 for 1. The proof went to Sonnet, while many short factual and coding requests went to Luna. This is a dated sample of choices, not the router's complete model pool or a routing rule you can rely on.
Method and evidence · Checked 2026-10-02
Method: 24 prompts, one run each, through OpenRouter on 2026-10-02, with max_tokens at 400. Answers with a fixed result were checked by string match; open-ended ones were not scored, so what Jev Router picked for them is recorded but not graded.