403 error on cloud API

[rog0978]

I am using the Tesla Cloud configuration, and I started to get 403 authentication error on 12-Jun:

"HTTPError('403 Client Error: forbidden, see https://developer.tesla.com/docs/fleet-api for url: https://owner-api.teslamotors.com/api/1/products'))."

Other integrations are seeing the same. For teslamate, this PR fixed the issue.

[spoonwzd]

Yup same

On pypowerwall startup:

06/14/2026 10:28:23 AM [pypowerwall.cloud.pypowerwall_cloud] [ERROR] Failed to retrieve sitelist - HTTPError('403 Client Error: forbidden, see https://developer.tesla.com/docs/fleet-api for url: https://owner-api.teslamotors.com/api/1/products')

Wasn't sure if it was just my code hammering it every 5 mins that caused the issue after failed attempts to set the reserve. Seems not.

Also lots of these:

06/14/2026 08:28:11 AM [pypowerwall.local.pypowerwall_local] [ERROR] Unexpected error connecting to Powerwall at https:///api/operation: Invalid URL 'https:///api/operation': No host supplied
06/14/2026 08:28:15 AM [pypowerwall.local.pypowerwall_local] [ERROR] Unexpected error connecting to Powerwall at https:///api/operation: Invalid URL 'https:///api/operation': No host supplied
06/14/2026 08:28:15 AM [pypowerwall.local.pypowerwall_local] [ERROR] Unexpected error connecting to Powerwall at https:///api/status: Invalid URL 'https:///api/status': No host supplied
06/14/2026 08:28:15 AM [proxy] [INFO] Control Command: Set Reserve to 5

[jasonacox-sam]

Thanks for the report and the TeslaMate reference — this is the same Tesla-side API change tracked in #317 and #318 (both closed).

What happened: Starting around June 12, Tesla's owner-api.teslamotors.com began enforcing HTTP/2 and TLS 1.3. The TeslaMate fix you linked (teslamate#5405) addresses exactly this — their HTTP client was using HTTP/1.1 and/or TLS 1.2 and getting 403'd.

Impact on pypowerwall: This affects cloud mode (TeslaPy), which uses owner-api.teslamotors.com for authentication and data polling. The underlying HTTP client needs to negotiate HTTP/2 + TLS 1.3 to comply with Tesla's new requirements.

Workaround for now: If you're able to switch to Fleet API mode, it uses a different OAuth endpoint (fleet-api.prd.na.vn.tesla.cn / regional) that isn't affected by this change. Several users in #318 confirmed that Fleet API mode works fine while cloud mode returns 403.

@jasonacox — flagging this as needing a code-level fix in the cloud module (or TeslaPy dependency) to enforce HTTP/2 + TLS 1.3 on owner-api connections, similar to what TeslaMate did in #5405.

— Sam ⚡

[jasonacox]

@jasonacox-sam it sounds like the code change won't really work, but we can still explore.

In the meantime, can you submit a PR to update the README.md to notify the community of the depreciation of the Tesla Owner API (cloud mode)?

[rog0978]

@jasonacox what is the reason for the code change not working?

[jasonacox-sam]

PR #326 adds deprecation notices to the README for Cloud Mode (Option 3): #326

Two callouts — a top-level banner and a section-level notice in the Cloud Mode setup — direct users to migrate to FleetAPI Cloud Mode (Option 2) and link back here for context.

— Sam ⚡

[jasonacox]

@rog0978 - have you been able to get it to work? I'm still getting 403. I'll dive deeper when I get a chance, unless Sam finds something else.

[goodoldme]

@jasonacox-sam @jasonacox I got the exact error while running tesla-history and got it working by following jasonacox/Powerwall-Dashboard#779 (comment)

pip install httpx h2
pip install --force-reinstall --no-deps git+https://github.com/jasonacox-sam/pypowerwall.git@fix/http2-auth

I ran tesla-history directly from the host. I am assuming tesla-history calls the same API endpoints as pypowerwall. Wondering if there's any difference running it directly on the host vs. running it on docker.

Just noticed that this will possibly require a docker change as well. It's mentioned in #317 (comment)

Does the docker change needs to be included in PR #324 to make this work?