Inside The Machine

← All business recipes

B-05Tool · assembly sheet

Local Jev

A decide-only model on your own machine. It answers typed questions about each item: pick one, give a score, or yes/no. It never writes.

Title blockbuild.ts

Part no.
B-05
Family
Creator ops
Runs from
8 GBParts run one at a time
Status
Partly tested
In
A JSON file of items and typed questions.
Out
Strict JSON answers, each marked [sure] or [UNSURE], plus a receipt.
Tools
Ollama
Note
Runs today. Tested in a cloud container on CPU with Qwen3.5 0.8B. Speed not measured.
1Claude's job

Claude does the few hard parts.

Planning, the hard writing and the final review. That takes few tokens.

Claude does
  • Writes the questions once.
  • Decides the items Local Jev marks [UNSURE].
Claude never sees

Everything runs against Ollama on your own computer. The tool refuses to talk to any other host. --dry-run shows exactly what the model would read.

2Local partsFig. 1 · Parts listmodels.json ·

Your machine does the bulk.

Find your memory. Each row names the model for each part. Free, private, on your computer.

The model for each part at each memory size. Parts run one at a time.models.json ·
No.MemoryDecider (Local Jev)
SZ-018 GB6 GB usableQwen3.5 4Bneeds 5.8 GBRuns, but tight
SZ-0216 GB12 GB usableQwen3.5 0.8Bneeds 2.6 GBRuns well
SZ-0324 GB18 GB usableQwen3.5 4Bneeds 5.8 GBRuns well
SZ-0432 GB24 GB usableQwen3.5 4Bneeds 5.8 GBRuns well
SZ-0548 GB36 GB usableGemma 4 12Bneeds 11.2 GBRuns well
SZ-0664 GB48 GB usableGemma 4 12Bneeds 11.2 GBRuns well
SZ-0796 GB72 GB usableQwen3.8 27Bneeds 22.7 GBRuns well
SZ-08128 GB96 GB usableQwen3.8 27Bneeds 22.7 GBRuns well
SZ-09192 GB144 GB usableQwen3.8 27Bneeds 22.7 GBRuns well
SZ-10256 GB192 GB usableQwen3.8 27Bneeds 22.7 GBRuns well
SZ-11512 GB384 GB usableQwen3.8 27Bneeds 22.7 GBRuns well
Decider (Local Jev)

Answers typed questions about each item: pick one, give a score, or yes/no. It never writes.

How we pick it: The smartest chat model with an Ollama download that needs no more than a third of your usable memory. Smaller models answer sooner, and a decider only picks a label. If no model is that small, the drafter's model decides too.

Fit labels

We count 75% of your memory as usable for AI. The rest is for the system and your apps. Runs well: the model fits in usable memory with at least 10% to spare. Runs, but tight: it fits, but less than 10% of usable memory is left. Close other apps first. Needs N GB+: it does not fit. N is the smallest memory size it fits in.

3Steps

6 steps.

Each step says who does it: you, Claude, your machine or a script.

  1. Claudewrite

    Write the questions

    One clear question each: choice, score or yes/no, with criteria in plain words.

  2. Your machineDecider (Local Jev)probe

    Read the odds

    One short call per question reads how likely each answer is.

  3. Your machineDecider (Local Jev)answer

    Answer in strict JSON

    A second call answers through Ollama structured outputs, so the reply must match the schema.

  4. A scriptmark

    Mark sure or UNSURE

    Choice and score: sure at 60% or more. Yes/no: yes at 90%+, no at 10% or less. If the two calls disagree, it is UNSURE.

  5. Youreview

    You decide the unsure ones

    The last line lists every UNSURE answer.

  6. Youshare

    Share your run (optional)

    At the end, the script shows exactly what it would upload: your chip, memory, the models, how long each stage took and the token counts. No prompts, no outputs, no file names. Say yes and it goes to our review queue. Nothing is public until we approve it.

4Make it yours

The step only you can do.

Before you publish
  • A verdict never triggers a delete, send or spend on its own.
5Receiptsapproved runs

What a run costs.

Claude tokens for this recipe and for doing it all in Claude, local minutes and dollars. Only from runs a person has approved.

Receipts

No runs yet
Claude tokens, this recipe
not run yet
Claude tokens, all in Claude
not run yet
Local minutes
not run yet
Claude dollars (estimate)
not run yet
Not run yet

Not run yetBlueprints now, numbers later. We run every recipe on a 128 GB Mac Studio from 10 Nov 2026, then publish the receipts.

6Run ittools/

Get it and run it.

You need Node 20 or newer and Ollama with one model. Read a script before you run it.

local-jev.mjsLocal Jev
mkdir -p itm-tools/local-jev itm-tools/local-intern && cd itm-tools
curl -fsSL https://insidethemachine.ai/tools/local-jev/local-jev.mjs -o local-jev/local-jev.mjs
curl -fsSL https://insidethemachine.ai/tools/itm-share.mjs -o itm-share.mjs
cd local-jev
ollama pull <your model>
node local-jev.mjs request.json --model <your model> --dry-run
node local-jev.mjs request.json --model <your model>
Read the script
7Share your run

Share your run. Only if you want to.

At the end, the script shows exactly what it would upload and asks first. A person reviews every run before it shows here.

It sends
  • Your chip, memory, graphics card and operating system
  • The recipe and the models you used
  • How long each stage took
  • Token counts: Claude in and out, local in and out
  • How many items came out
  • Tool versions
  • A fingerprint (SHA-256) of the log and up to 15 cleaned log lines
  • A handle you pick (not your name or email)
It never sends
  • Your prompts or the model's answers
  • Your files, file names or folder paths
  • Emails, links, keys or IP addresses (the server rejects any upload that has one)
  • Your IP address in plain form: we keep a scrambled copy for rate limits only

Agents can share with the submit_run tool of https://insidethemachine.ai/mcp. Privacy

Agent view · text/markdown · or send Accept: text/markdown
# Tool: Local Jev

URL: https://insidethemachine.ai/build/local-jev
Part: B-05. Family: Creator ops. Status: Runs today (CPU test, speed not measured).
Runs today. Tested in a cloud container on CPU with Qwen3.5 0.8B. Speed not measured.

A decide-only model on your own machine. It answers typed questions about each item: pick one, give a score, or yes/no. It never writes.

- Input: A JSON file of items and typed questions.
- Output: Strict JSON answers, each marked [sure] or [UNSURE], plus a receipt.
- Runs from: 8 GB of memory.
- Tools: ollama (https://insidethemachine.ai/setups/ollama.md)

## Claude's job

- Writes the questions once.
- Decides the items Local Jev marks [UNSURE].

What Claude never sees: Everything runs against Ollama on your own computer. The tool refuses to talk to any other host. --dry-run shows exactly what the model would read.

## Local parts per memory size

Data: models.json 2026-10-15, speech.json 2026-10-15, images.json 2026-10-15.
| memory GB | Decider (Local Jev) |
|---|---|
| 8 | Qwen3.5 4B (needs 5.8 GB, runs, but tight) |
| 16 | Qwen3.5 0.8B (needs 2.6 GB, runs well) |
| 24 | Qwen3.5 4B (needs 5.8 GB, runs well) |
| 32 | Qwen3.5 4B (needs 5.8 GB, runs well) |
| 48 | Gemma 4 12B (needs 11.2 GB, runs well) |
| 64 | Gemma 4 12B (needs 11.2 GB, runs well) |
| 96 | Qwen3.8 27B (needs 22.7 GB, runs well) |
| 128 | Qwen3.8 27B (needs 22.7 GB, runs well) |
| 192 | Qwen3.8 27B (needs 22.7 GB, runs well) |
| 256 | Qwen3.8 27B (needs 22.7 GB, runs well) |
| 512 | Qwen3.8 27B (needs 22.7 GB, runs well) | 

How each part is picked:
- Decider (Local Jev): Answers typed questions about each item: pick one, give a score, or yes/no. It never writes. Rule: The smartest chat model with an Ollama download that needs no more than a third of your usable memory. Smaller models answer sooner, and a decider only picks a label. If no model is that small, the drafter's model decides too. (models.json)
- Fit: We count 75% of your memory as usable for AI. The rest is for the system and your apps. Runs well: the model fits in usable memory with at least 10% to spare. Runs, but tight: it fits, but less than 10% of usable memory is left. Close other apps first. Needs N GB+: it does not fit. N is the smallest memory size it fits in. Size unknown: we do not know how much memory it needs.

## Steps

1. **Write the questions** (Claude; stage id `write`): One clear question each: choice, score or yes/no, with criteria in plain words.
2. **Read the odds** (Your machine, Decider (Local Jev); stage id `probe`): One short call per question reads how likely each answer is.
3. **Answer in strict JSON** (Your machine, Decider (Local Jev); stage id `answer`): A second call answers through Ollama structured outputs, so the reply must match the schema.
4. **Mark sure or UNSURE** (A script; stage id `mark`): Choice and score: sure at 60% or more. Yes/no: yes at 90%+, no at 10% or less. If the two calls disagree, it is UNSURE.
5. **You decide the unsure ones** (You; stage id `review`): The last line lists every UNSURE answer.
6. **Share your run (optional)** (You; stage id `share`): At the end, the script shows exactly what it would upload: your chip, memory, the models, how long each stage took and the token counts. No prompts, no outputs, no file names. Say yes and it goes to our review queue. Nothing is public until we approve it.

## Make it yours

- A verdict never triggers a delete, send or spend on its own.

## Receipts

- Claude tokens, this recipe: not run yet
- Claude tokens, all in Claude: not run yet
- Local minutes: not run yet
- Claude dollars (estimate): not run yet

No approved runs yet. Blueprints now, numbers later. We run every recipe on a 128 GB Mac Studio from 10 Nov 2026, then publish the receipts.

## Run it

Source: https://insidethemachine.ai/tools/local-jev/local-jev.mjs

```sh
mkdir -p itm-tools/local-jev itm-tools/local-intern && cd itm-tools
curl -fsSL https://insidethemachine.ai/tools/local-jev/local-jev.mjs -o local-jev/local-jev.mjs
curl -fsSL https://insidethemachine.ai/tools/itm-share.mjs -o itm-share.mjs
cd local-jev
ollama pull <your model>
node local-jev.mjs request.json --model <your model> --dry-run
node local-jev.mjs request.json --model <your model>
```

## Share your run

Opt-in. The script shows exactly what it would upload and asks first. It sends: Your chip, memory, graphics card and operating system; The recipe and the models you used; How long each stage took; Token counts: Claude in and out, local in and out; How many items came out; Tool versions; A fingerprint (SHA-256) of the log and up to 15 cleaned log lines; A handle you pick (not your name or email). Never: Your prompts or the model's answers; Your files, file names or folder paths; Emails, links, keys or IP addresses (the server rejects any upload that has one); Your IP address in plain form: we keep a scrambled copy for rate limits only. Nothing is public until a person approves it. Agents can submit with the `submit_run` tool of https://insidethemachine.ai/mcp.

Privacy: https://insidethemachine.ai/privacy.md