How to make teslamate work without additional costs

[kirksigi]

Hi everyone,

First off, a big thank you for TeslaMate. I've been running it since day one of owning my Tesla, and I love it.
Unfortunately the Owner API has stopped working, and migrating to the Fleet API turned out to be quite an adventure — even with AI lending a hand.
My goal: get TeslaMate running again without any extra cost. Tesla's Fleet API is technically paid, but as long as usage stays within the free allowance (under ~10 EUR), it costs nothing. Easier said than done, though — I'm stuck choosing between polling and streaming:
Polling: Accurate, but absurdly expensive. A single 100 km drive already burns through roughly 4 EUR in API usage.
Streaming: Inaccurate. Short drives drop out entirely. I haven't tried streaming via myTeslaMate — and since it's paid anyway, it's off the table for me.

For the first time my teslamate is not collecting data anymore. I had to switch it off.

What's your experience been? Has anyone gotten TeslaMate working properly on the Fleet API, and what does your usage look like? It seems like the high polling rate while driving/charging is the real bottleneck here.
Here are the environment variables from my docker file (no streaming = clean data, but high cost). In the TeslaMate settings, stream is set to off:

      - ENCRYPTION_KEY=XXXX
      - DATABASE_USER=teslamate
      - DATABASE_PASS=teslamate
      - DATABASE_NAME=teslamate
      - DATABASE_HOST=database
      - MQTT_HOST=mosquitto
      - MQTT_USERNAME=XXXX
      - MQTT_PASSWORD=XXXX
      - TZ=Europe/Vienna
      - TESLA_API_HOST=https://fleet-api.prd.eu.vn.cloud.tesla.com
      - TESLA_AUTH_HOST=https://auth.tesla.com
      - TESLA_AUTH_PATH=/oauth2/v3
      - TESLA_AUTH_CLIENT_ID=XXXX
      - TESLA_WSS_HOST=wss://streaming.myteslamate.com
      - TESLA_WSS_TLS_ACCEPT_INVALID_CERTS=true
      - TESLA_WSS_USE_VIN=true

If you wonder: Mosquitto runs in a sepparate container.

[ngardiner]

I realise that what I'm about to say is only marginally useful, but I have been wrestling with this myself and I will say that yes, it is absolutely without a doubt 100% doable, but without some work to align with the fleet telemetry server (and I just wouldn't do it with polling, as you say it is absurdly expensive) it won't just work out of the box

The reason you found that short drives didn't work with Fleet Telemetry is that the core engine still works on a polling mechanism, and the only way to do that efficiently is to dial the polling WAY down, which then makes the responsiveness to drive state much less accurate

What I have done, and I have no interest in open sourcing it because frankly it's entirely vibe coded "proof of concept" code is to connect the myTeslamate API layer with Fleet Telemetry server, and then add in pseudo state for the standard polling API endpoint based on streaming state, ie it will reflect driving, charging etc based on the streaming data. because streaming is delta only, no streamed telemetry for 15m = sleep state. I ended up sending all wake commands to a no-op, and dialled TeslaMate's API polling down to seconds because it's now zero-cost given it is completely inferred from streaming state

I ran into a ton of corner cases and challenges, I'll have a small bill from Tesla this month for my poking and prodding, but I at least feel informed about what is and isn't possible and I can say that as long as:

• You're able to host a fleet telemetry server
• You aren't running other tesla integrations sending expensive wake messages or commands
• You explicitly block or reconfigure all the behaviours that lead to cost which are inherent to the current design of TM, or build middleware to block them out

You can absolutely drive a ton with 1 second telemetry frequency (which is 1/3 of the native API rate) and stay within the $10 allocation, as you are ONLY incurring the telemetry streaming cost, which is $0.0001 per signal (data point)

But you wouldn't be able to get there today with the current tunables - it needs a change from polling based to passive telemetry collection - you also lose the "power" value during drive, 1/3 less resolution and probably will exceed the $10 allocation if you are driving WAY more than the average driver, have multiple vheciles, etc. given it's based on activity.

One last learning that I'll impart on people because I found it annoying but am now aware - if you do limit API costs (which you should) and for example wake up to find that a misconfigured setting meant the API got a hammering overnight and thankfully you had the foresight to cut it off - just remember that the way Tesla limit inbound telemetry when you exceed spend caps is by deleting the telemetry configuration from the vehicle, so you need to send it back to the car again, simply increasing spend limit won't bring everything back again. It's been a learning process (and to be quite honest it's been very Tesla. not at all developer friendly)

[fals3illusion]

I am currently tuning this too.... once my teslamate stopped working from the old ownerapi i updated my container and the first day i drove a lot and almost burned though my free $10. i got an email about it and looked into it currently these are my variables i am tuning and will post back once i get a dialed in compromise for reliability for short drives like you mentioned, and good data collection.

  - POLLING_DRIVING_INTERVAL=45
  - POLLING_CHARGING_INTERVAL=600
  - POLLING_ONLINE_INTERVAL=600
  - POLLING_ASLEEP_INTERVAL=300
  - POLLING_DEFAULT_INTERVAL=600
  - POLLING_MINIMUM_INTERVAL=45

[lotharbach]

Owner API did not stop working, there was a (technically quite small) change on tesla's server that is fixed in the latest teslamate release.
If it isn't working for you on the latest release, it's very likely an issue with your setup and you should try again.

[ngardiner]

Some of us see the writing on the wall re the death of the owner api now that it has been moved to the same ingress as the fleet API

Taking the pain now is going to make it easier when it happens

[lotharbach]

But it's just not something one needs to do right now, and not how the question was framed. Today, the best way to use Teslamate without any costs is Owner API. It's been about 3 years that people predicted the imminent death of the owner api.

I just fear for your struggles to be for nothing after all, if a really targeted shutdown finally comes the entire project will of course move and that will then look different than any of the current hacks (for the streaming/telemetry part, which is just mandatory to work)

[fals3illusion]

Sorry, I though the owner API was finally phased out... couldn't get it going and worked with Claude and guess it lied to me and was directing only to fleet-api ha ha . Turns out i was trying to get auth codes from fleet api to work on the owner api host, then got fleet api working and stuck with it once i updated everything to the new Teslamate version.

Once i got the correct owner api auth codes and host, etc everything worked like before.

Thanks for confirming that.... guess i now know how to get a telemetry server and the fleet api working when the time comes ha ha

[fals3illusion]

Hi everyone,

First off, a big thank you for TeslaMate. I've been running it since day one of owning my Tesla, and I love it. Unfortunately the Owner API has stopped working, and migrating to the Fleet API turned out to be quite an adventure — even with AI lending a hand. My goal: get TeslaMate running again without any extra cost. Tesla's Fleet API is technically paid, but as long as usage stays within the free allowance (under ~10 EUR), it costs nothing. Easier said than done, though — I'm stuck choosing between polling and streaming: Polling: Accurate, but absurdly expensive. A single 100 km drive already burns through roughly 4 EUR in API usage. Streaming: Inaccurate. Short drives drop out entirely. I haven't tried streaming via myTeslaMate — and since it's paid anyway, it's off the table for me.

For the first time my teslamate is not collecting data anymore. I had to switch it off.

What's your experience been? Has anyone gotten TeslaMate working properly on the Fleet API, and what does your usage look like? It seems like the high polling rate while driving/charging is the real bottleneck here. Here are the environment variables from my docker file (no streaming = clean data, but high cost). In the TeslaMate settings, stream is set to off:

      - ENCRYPTION_KEY=XXXX
      - DATABASE_USER=teslamate
      - DATABASE_PASS=teslamate
      - DATABASE_NAME=teslamate
      - DATABASE_HOST=database
      - MQTT_HOST=mosquitto
      - MQTT_USERNAME=XXXX
      - MQTT_PASSWORD=XXXX
      - TZ=Europe/Vienna
      - TESLA_API_HOST=https://fleet-api.prd.eu.vn.cloud.tesla.com
      - TESLA_AUTH_HOST=https://auth.tesla.com
      - TESLA_AUTH_PATH=/oauth2/v3
      - TESLA_AUTH_CLIENT_ID=XXXX
      - TESLA_WSS_HOST=wss://streaming.myteslamate.com
      - TESLA_WSS_TLS_ACCEPT_INVALID_CERTS=true
      - TESLA_WSS_USE_VIN=true

If you wonder: Mosquitto runs in a sepparate container.

FYI, these are the settings I used to get my Owner API working with the new Teslamate v4.0.1

  - ENCRYPTION_KEY=${ENCRYPTION_KEY}
  - DATABASE_USER=teslamate
  - DATABASE_PASS=${DATABASE_PASS}
  - DATABASE_NAME=teslamate
  - DATABASE_HOST=${DATABASE_HOST}
  - MQTT_HOST=${MQTT_HOST}
  - MQTT_USERNAME=${MQTT_USERNAME}
  - MQTT_PASSWORD=${MQTT_PASSWORD}
  - TZ=America/Chicago
  - TESLA_API_HOST=https://owner-api.teslamotors.com
  - TESLA_AUTH_HOST=https://auth.tesla.com
  - TESLA_AUTH_PATH=/oauth2/v3
  - TESLA_AUTH_CLIENT_ID=ownerapi

  
  then get an Owner API auth and refresh code from this Windows tesla-auth program
  https://github.com/adriankumpf/tesla_auth/releases/download/v0.13.0/tesla_auth-x86_64-pc-windows-msvc.zip
  
  That got me working back to normal.

[kirksigi]

Wow. That really did it. Thanks man.
Lets see how long this will last.

[brianmay]

TESLA_STREAMING_DISABLE=true

There shouldn't be any need to disable streaming.

[fals3illusion]

correct, whoops, i forgot owners api you want streaming enabled. i commented it out right after this post and forgot to edit this thread good catch

[jlestel]

Since the theme here is running TeslaMate at zero cost, one more option worth putting on the table: MyTeslaMate just shipped a free tier aimed exactly at this: keep your self-hosted TeslaMate and connect the Tesla side through our proxy at no cost.

How it stays free (try it here):

• A degraded Fleet Telemetry stream provides the live data: fewer fields and longer intervals than the paid tier, but enough for the essentials:
• Drives — position, speed, odometer, SOC, range
• Charges — charge state, AC/DC power, energy added (kWh), SOC
• Battery level

Because it combines the cached vehicle_data with the stream, it doesn't drop short drives the way pure streaming does.

What stays on the paid API plan: climate, TPMS, doors/windows, sentry, ~5× more signals at higher precision, and vehicle commands.

Setup is just a handful of env vars on your existing TeslaMate (TESLA_API_HOST, TOKEN, TESLA_WSS_HOST, …): you'll find the exact values on the fleet page, and they're also documented on TeslaMate. Nothing else changes.

Would the degraded telemetry set be enough for your use? We've been actively growing the free field set based on this kind of feedback (recently added battery level, detailed charge state, and charge energy), so I'm curious what you'd consider missing.

[kirksigi]

Thanks. I will give it a shot.