Issues with TeslaMate and new Fleet API Tokens

[Pier276]

I've spent an excessive number of hours attempting to install TeslaMate with Docker on a Linux Mint then Windows 11 box to access the Tesla API for my 2025 Model 3 Highland. I'm continuing to have issues with the tokens being recognized when logging into TeslaMate at localhost:4000.

I has become sufficiently frustrating that I resorted to asking Gemini then Grok then Gemini and finally ChatGPT. Gemini and Grok sent me down numerous rabbit holes that didn't work. ChatGPT got me the closest, but finally stated that "I'm doing everything right" (in ChatGPT's mind LOL) but that TeslaMate was built for TeslaMate was built around the old Owner API + streaming and
Tesla is forcing migration to Fleet API + Telemetry. This is the cause of my problem.

I don't want to pay SaaS to access MY data on MY car. I'm doing this as a hobbiest for fun. ChatGPT is recommending I can this project with TeslaMate and move to Home Assistant.

Just looking for some sanity here.

TIA

[Pier276]

UPDATE: I apology if this is redundant. I did search and didn't find anything. AI tools indicate they scrubbed GitHub and there are reports that to "fix" this, it would require a rewrite of TeslaMate.

[brianmay]

Not really sure how to help, because not sure what you are trying to do, and what you have already tried...

AFAIK the old owners API still works fine, and it is what is recommended to use with Teslamate.

Did you see the FAQ? In particular the first question? https://docs.teslamate.org/docs/faq/

And it is still working fine for me, no sign of any forcing migration really.

The new fleet telemetry API does have some benefits. It is technically better, no need to poll the car frequently but not too frequent and prevent it doing to sleep. But it does have some downsides too. At least last I looked.

For me the biggest show stopper is some of the conditions you have to agree to are very dodgy, so much that I have not been interested in agreeing to them.

And I think it is non-trivial - at least last time I checked - to distribute and configure open source software that directly uses the new APIs.

And of course, it is no longer free. There is free usage tier, but accidentally go over the limits and you will get charged.

[Pier276]

Thanks, Brian. I'm was just trying to connect TeslaMate to my car to monitor the performance and track the car. I don't really need to control the car as I have the Tesla mobile app and have installed S3XY Commander with it's app to control most of what I want to control.

I was attempting to run TeslaMate on a local Linux box running in a Docker container using a registered domain with DNS and Cloudshare to the Tesla API. I have everything installed virtual key installed on the car so registered as a Tesla API developer. BUT, I can't log into the TeslaMate portal at localhost:4000. I create the keys. Paste them into the TeslaMate portal to login and I keep getting errors on invalid keys.

I was using the Fleet API because it seemed that was the push by Tesla and that the Owner ID was in the plans for being replaced by it. I have been getting mixed information regarding the use of Owner API tokens continuing and that TeslaMate was developed for the Owner API and several people are having problems interfacing with the Fleet API because TeslaMate wasn't developed for it.

Additionally, I've been reading that I should just incorporate the access to my Tesla Model 3 Highland through Home Assistant.

Sorry, for the long answer to your response but I wanted to be complete.

[frederik-smrthme]

My thoughts (partially off-topic) on this issue after 2months journey through data-logging with teslamate & teslalogger .

I ran thhrough all kind of setups. Running teslamate with fleet api (using myteslamate-proxy) and also alternetively running teslalogger with fleet-api and own telemetry-server (because i did not want to pay for this service). But Teslamate is not made for fleet-api
(Edit: teslalogger should work with selfhosted telemetry-server. But in my case it did not). However, maybe its somehow possible to get those things work, but never forget what could happen if updates should be made in future. So it will most-likely end up in too much maintenance to have a reliable system.

Well, as I could not take a decision, I ended up having both loggers running in parallel on a server with a dedicated domain. Teslalogger witth fleet-api and teslamate with owners-api to get a feeling for the differences in accuracy and user-friendlyness and the day will come where i prefer one of both.
Whilst Teslalogger fully intergrates elevation-data into the drives, this is somehow missing in teslamate. I therefore added a separeate docker-stack that scans every single trip of teslamate-drives and adds elevation-data to the positions.

My outcome so far. If you definitely want to use fleet-api there are two options that will work on a dedicated server:

1. use teslamate with myteslamate's fleet api-proxy (at no cost!)

UPSIDE:

• nicer dashboards than teslalogger,
• user-firendy (esp. for non-tecchies)

DOWNSIDE:

• akaik the fleet api data is converted to owers-api format (for teslamate).
• Sometimes I faced missing data caused by crappy data-proxy setup within myteslamate. This can crash consumption-statitstics and only be corrected by manually fixing database-entries.

2. use teslalogger and pay 3€ / month for fleet-api usage

UPSIDE:

• data-driven,
• able to integrate scan-my-tesla and have better view on battery- and single-cell data.

DOWNSIDE:

• not able to use of Matedroid with teslalogger (android-app).
• Grafana-dashboard is not a fancy as Teslamate.

[bassmaster187]

and teslalogger somehow blocks self-hosted telemetry-server if you don't pay for it.

That's not true. We have many advanced users using Teslalogger with own Tesla Telemetry Server. 4 of them are in our developer team and are happy to help you. I accepted every pull request to make it as easy as possible to use an own server.
You can find many infos here:
bassmaster187/TeslaLogger#1401

[frederik-smrthme]

thx for your reply and clarification, Christian. Then ist my fault. I edited my post to make this clear. However, after spending many hours on this matter, meanwhile I perfer to pay for this service. Maintaining telemetry-server and infrastructure for one single car does not really make sense.

[bassmaster187]

Yes, I admit the whole process Tesla created for Fleet Telemetry is quite complex. I’m not sure if it’s still as complicated as it was two years ago—for example, back then I couldn’t create a developer account without a VAT number.

But if you enjoy solving challenging problems, running your own free Fleet Telemetry server is definitely possible. The most difficult part is getting a developer account and setting up all the certificate-related requirements on Tesla’s telemetry server. Once that’s done, it’s just a single setting in Teslalogger to process the messages.

And yes, I could setup an own free Telemetry server for myself, but I prefer to pay for the service, so I have to take care just for one Telemetry Server and not for two.

[Pier276]

I just want to add that I appreciate this discussion. Thank you. For me, honestly, at this time I don’t see the value, paid or unpaid. It was to be a fun learning project.

I’ve learned the effort to kludge it together with these tools isn’t worth the skin. But, the nerd in me wants a tech project. I have a myriad of smart home devices running on their own separate cloud platforms. Lately post acquisition our new ISP has become quite unstable causing havoc with these devices. I’m motivated to bring all of this back in-house.

I’m going to attempt a new Home Assistant install on Ubuntu Server and Docker. At least I have the benefit of home automation consolidation. At the end of that I’ll look into the Tesla integration as part.

Thanks again.

[Pier276]

Thank you for taking the time for this detailed response.

I started my project for fun and nothing more. The car is more of a computer on wheels and with the API provides access to more data than any car I had owned previously. In my past I have developed on multiple platforms in multiple programming languages so relatively confident I could accomplish it somewhat easily. I was unaware of the details of limitations with TeslaMate limitations with the Fleet API.

I really don’t need yet another SaaS so I may consider your option #1 in a few weeks. But TBH, it seems similar to the ICE design where patch after patch to a deficient system ultimately results in an inferior product. My TeslaMate is a patch that with one change in the chain can become a maintenance effort that is greater than the benefit of the solution.

Again, thank you for the education. You have helped save what hair I have left.

[AMX-Claw]

Hit the same wall mid-June (TeslaMate's built-in refresh sends client_id="ownerapi" which Fleet API accounts reject → tokens go stale → everything 401s, plus the v4.0.1 crash loop when the streaming websocket drops).

Got a stock TeslaMate v4.0.1 container fully working again on Fleet API with three small config-only pieces — an auth proxy that rewrites client_id to your own Fleet API app id, a 6-hourly token rotation cron that injects fresh tokens into TeslaMate's encrypted store, and a websocket proxy that absorbs upstream streaming drops so TeslaMate stops crash-looping.

Running for ~2 weeks on a Pi 4; polling and streaming drive detection both working. Packaged it up in case it saves someone the digging: https://github.com/AMX-Claw/teslamate-fleet-proxy

(You still need your own Fleet API developer app/client_id — that part nobody can ship for you, sadly.)