You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Error from provider (Console): OpenCode's free tier can only be used from within OpenCode #49433
The error message is slightly different ("can only be used from within OpenCode" vs "1.17.0 or newer is required"), but both relate to the free tier Console provider check. If your issue is distinct, please clarify what client/interface you are using (TUI, Desktop, or API) and whether you have any custom provider configuration.
Root cause: the free-tier gate reads the client version from User-Agent; non-release version strings are rejected
Environment — Arch Linux, two builds on the same machine, identical config ({"$schema":...} minimal), network and proxy, anonymous (no OpenCode credential):
opencode1.18.31 (official release build) → ✅ works
1.18.29 (official release) → ✅ works
Repro: opencode run -m opencode/mimo-v2.5-free "hi" → Error from provider (Console): OpenCode's free tier can only be used from within OpenCode (deterministic, fails in under a second)
Evidence:
Captured both clients' outbound requests against a local mock (provider.opencode.options.baseURL override). Headers identical except the version inside User-Agent:
Authorization: Bearer public and all x-opencode-* headers were the same.
Decisive test: in a local copy of the failing binary, patched only the version literal (1.18.31.r12.g88c6c7a → 1.18.31, same byte length, nothing else) → free tier works immediately. The version string is the discriminator.
Implication: any development/nightly/-git build whose version is a git-describe string (<tag>.rN.gSHA) is not valid semver and gets rejected. The 1.17.0 or newer is required variant (#49430, #49431) looks like the same version-parsing gate with a different message.
Suggested fixes (either side): report a sanitized semver in User-Agent for non-release builds (keep the full git-describe string in --version/logs), or make the server tolerate such strings / gate on an explicit client marker instead of the version.
A third-party client (pi) hits the same gate, and the User-Agent version is only one of two checks
Thanks @LouisLau-art for isolating the User-Agent check. I hit the same error from a different client and reproduced it on purpose. This adds a second, independent condition the gate enforces.
Context
pi is a coding agent with a built-in opencode provider. It sends x-opencode-client: pi for attribution. Since 2026-09-17 the free tier returns for every free model:
OpenAI API error (403): {"type":"FreeTierError","message":"Error from provider (Console): OpenCode's free tier can only be used from within OpenCode"}
The error appears on two hosts (macOS and Windows), with the same key and the same pi version, and without the OpenCode CLI involved. That is why I looked at the request fingerprint rather than at the account.
Method
I sent direct requests to https://opencode.ai/zen/v1/responses with Authorization: Bearer public (anonymous free tier) and a minimal JSON body, then varied one header at a time.
Finding 1 — User-Agent must be opencode/<recent semver>
User-Agent
Result
opencode/1.18.31
200
opencode/latest/1.18.31/cli
200
opencode/999.0.0
200
OpenCode/1.18.31
200
opencode/0.0.0
426
opencode/latest
403
opencode/
403
opencode (no slash)
403
pi (darwin 25.6.0; arm64)
403
curl/8.7.1
403
OpenAI/JS 6.40.0
403
header absent
403
This matches #49430/#49431 (1.17.0 or newer is required) and the git-describe rejection. The trailing SDK tags (ai-sdk/provider-utils/... runtime/bun/...) are not required: plain opencode/1.18.31 passes, and OpenCode/1.18.31 passes, so the match is case-insensitive.
Finding 2 — x-opencode-session must be an OpenCode session ID
With User-Agent: opencode/1.18.31 fixed, the session header is a second independent gate. The accepted format is ses_ + 12 hex characters + 14 base62 characters, i.e. what Identifier.create produces in packages/opencode/src/id/id.ts.
The 12-hex time component is not checked for recency. Timestamps from now back to one year ago pass; an all-A component (which decodes to ~1971) fails. So the gate looks for a plausible timestamp, not an exact one.
x-opencode-client is not validated. cli and pi both pass with a valid session.
x-opencode-request and x-opencode-project are not required.
The body is irrelevant: a full agent payload and a minimal {"model","input","stream"} payload behave the same.
The gate is not key-based: Bearer public and a real account key both pass with a valid fingerprint.
So a request to the free tier needs both an opencode/-style User-Agentand a ses_-formatted session ID. A third-party client that reports an honest User-Agent (as pi does) is rejected twice: once on the UA and once on its own UUID session.
Request
Could the OpenCode team make the free-tier client requirements explicit in the docs? Concretely:
Is the free tier intended to be limited to the official OpenCode client? If known third-party clients are supported, what should they send? pi already sends x-opencode-client: pi, which suggests third-party attribution was intended.
Can you document the accepted User-Agent form and the required x-opencode-session format in one place, so client authors can comply instead of reverse-engineering the edge service? A 1.17.0 or newer minimum and a session-ID requirement would be enough.
If the intent is official-client-only, please state that, and say whether an explicit client marker (rather than a spoofable UA/version and session format) is planned.
I am happy to test any documented header contract and to adjust pi accordingly.
same issue on big pickle, it specifically happens when it goes to perform compaction. Haven't changed anything in weeks, been using the same version and model and now suddenly today this happens
same issue on big pickle, it specifically happens when it goes to perform compaction. Haven't changed anything in weeks, been using the same version and model and now suddenly today this happens
For me, in the same session, i tried everything with big pickle, and nothing worked, still got the same error, and some other sessions simply work
opencode run --agent plan --model opencode/big-pickle still returns FreeTierError" / HTTP 403 (log id 75c7ff8e-9f06-40b3-9ab7-d8d5668c8375), while --agent build` works fine.
Since the only difference between the two requests is the agent toolset (plan runs without edit/write/bash in my config), this strongly suggests the gateway is inspecting the request toolset rather than the client or session. The documented read-only plan agent therefore cannot work on the free tier at all.
The Plan mode agent is working for me now but my read only Q/A agent that I created with markdown is triggering the "OpenCode's free tier can only be used..." error, so you might be on to something.
Thanks for investigating this. We can confirm a permission-dependent reproduction on Linux with the official npm CLI 1.18.32, model opencode/big-pickle, and a configured Zen API credential.
Failing command in an empty project:
OPENCODE_PERMISSION='{"*":"deny"}' opencode run --pure --agent build --format json --model opencode/big-pickle 'Do not use tools or access files. Reply exactly BIG_PICKLE_OK.'
This returns HTTP 403 / FreeTierError: "OpenCode's free tier can only be used from within OpenCode". We also observed the same error on 1.18.31. Failed request log ID: bcc8d6a8-aa88-48f0-9152-2b9cb5f08e49.
Using approval-gated tools instead of removing them resolves it here:
OPENCODE_PERMISSION='{"*":"ask","external_directory":"deny"}' opencode run --pure --agent build --format json --model opencode/big-pickle 'Do not use any tools or access files. Reply exactly BIG_PICKLE_OK.'
Result: BIG_PICKLE_OK, exit 0. Confirmed again after persisting permission: {"*":"ask", ...} in the global config and removing the environment override. Both successful requests reported cost 0, and neither executed a tool. No client/header spoofing, custom endpoint, or paid-model fallback was used.
Could the free-tier validation support legitimate deny/reduced-toolset configurations, or at least return an actionable error explaining the incompatibility? The current message led us to investigate account authentication and client versions unnecessarily. Happy to provide additional sanitized details. Thank you!
Custom agents with permission: bash: deny reliably get:
Error from provider (Console): OpenCode's free tier can only be used from within OpenCode
The same agent with bash: ask (or without a bash permission) succeeds
immediately. Built-in plan / build succeed. This looks like the free-tier
client gate misclassifies requests when bash is fully denied — not a missing
User-Agent/session header from an external harness.
Controlled A/B/A/B (same binary, same model, ~2026-09-23)
opencode run --agent chat --model opencode/mimo-v2.6-flash-free "Reply with exactly: HI"# → Error: OpenCode's free tier can only be used from within OpenCode
Change to bash: ask → succeeds. Pattern form also fails:
permission:
bash:
"*": deny
Other single-key tests with bash omitted all succeeded on Zen free: edit: deny, write: deny, lsp: deny, todowrite: deny, tavily_: ask, temperature: 0.5.
Cross-agent: a throwaway custom agent with only bash: deny also failed; a
minimal custom agent without a bash permission succeeded. So it is not
specific to the agent name chat.
Other observations
Title agent on the same host: gpt-5.4-nano → Model access is disabled (separate issue).
Workaround: use bash: ask (or omit bash) on agents that must use Zen free models.
Timeline matches the ~Sep 17 free-tier gate tighten; last successful chat + Zen free on this machine was 2026-09-08 (before the gate).
Environment
opencode --version: 1.18.32
OS: Linux
Plugins: claude-mem (no chat.headers hooks found)
Happy to run more header-level tests if maintainers need them.
Confirmed the workaround on Windows, Chocolatey 1.18.32: changing the agent's permission.bash from deny to ask resolves the 403 for the plan agent with opencode/big-pickle (verified both via CLI opencode run --agent plan returning the expected reply at cost 0, and in interactive plan-mode sessions). edit/write remain deny, so the agent is still effectively read-only — every execution prompts for approval.
Thanks to @orangeguitar for isolating the trigger to bash: deny, and to @sekedus, @joshportman and @gainfully4302 for the reproductions. The current error message (OpenCode's free tier can only be used from within OpenCode) really sends you down the wrong path (account/auth/client version) — an actionable message about the toolset requirement would save a lot of debugging. Thanks!
Adding a precise bisect in case it helps, since you asked about custom setups.
Setup: opencode 1.18.31 (official npm release, Linux/WSL2). Logged into Zen + Go. opencode is driven headlessly (opencode serve / opencode run) by an orchestrator. The worker agents are custom primary agents that start from "*": deny and only allow MCP tools that operate inside a sandbox container. Built-in bash/read are denied on purpose, so an automated worker can never touch the host.
Repro: a minimal custom agent and opencode run --agent <name> -m opencode/mimo-v2.6-flash-free "Reply with just OK", changing only one permission line each time:
agent permission
result
(none)
OK
bash: deny
403 "free tier can only be used from within OpenCode"
read: deny
403
edit / glob / grep / webfetch / task: deny
OK
"*": deny, bash: allow
403
"*": deny, bash: ask, read: ask
OK
So the gate seems to require bothbash and read to be present in the request's tool list. read: deny alone triggers it too, which I don't think has been reported here yet. ask passes because the tool stays in the list.
The same agents keep working on paid models: 187 completed Go runs since 2026-09-19, 0 rejected. Every free model is rejected; before 2026-09-19 the same agents had 459 completed free runs.
Restricting bash/read is a safety choice for unattended runs, not a sign of a different harness. It would be great if denying those tools didn't count as "not OpenCode". Happy to provide more details if useful.
I just started getting this error too. I’m not sure what’s causing it, it started happening randomly.
Happy to provide more information needed to debug this.
macOS 26.6.2
OpenCode 1.18.34 (updated from 1.18.32 to see if that would fix it, but it didn’t)
Nothing custom.
I was using MiMo-V2.6-Flash. Tried to switch to Space Bunny but same thing.
Me too. "Error: Error from provider (Console): OpenCode's free tier can only be used from within OpenCode".
Had also this error some weeks ago. Resolved by upgrading opencode.
I just upgraded opencode on my macos (macports: version opencode @2.0.21 llm/opencode) -- previously 1.18.31 iirc. But this time the error stays.
Thanks for the report! OpenCode 2.0 is now out and development is focused on V2, so we're closing issues filed against OpenCode 1. Please upgrade (opencode upgrade --major, details: https://anoma.ly/notes/opencode-2-0), and if this is still relevant on 2.0, reopen this issue or open a new one with your 2.0 version.
Activity
github-actions commentedon Sep 17, 2026
This issue might be a duplicate of existing issues. Please check:
The error message is slightly different ("can only be used from within OpenCode" vs "1.17.0 or newer is required"), but both relate to the free tier Console provider check. If your issue is distinct, please clarify what client/interface you are using (TUI, Desktop, or API) and whether you have any custom provider configuration.
LouisLau-art commentedon Sep 17, 2026
Root cause: the free-tier gate reads the client version from
User-Agent; non-release version strings are rejectedEnvironment — Arch Linux, two builds on the same machine, identical config (
{"$schema":...}minimal), network and proxy, anonymous (no OpenCode credential):opencode-git1.18.31.r12.g88c6c7a (archlinuxcn) → ❌ failsopencode1.18.31 (official release build) → ✅ works1.18.29(official release) → ✅ worksRepro:
opencode run -m opencode/mimo-v2.5-free "hi"→Error from provider (Console): OpenCode's free tier can only be used from within OpenCode(deterministic, fails in under a second)Evidence:
Captured both clients' outbound requests against a local mock (
provider.opencode.options.baseURLoverride). Headers identical except the version insideUser-Agent:opencode/1.18.29 ai-sdk/provider-utils/4.0.23 runtime/bun/1.4.0opencode/1.18.31.r12.g88c6c7a ai-sdk/provider-utils/4.0.23 runtime/bun/1.4.2Authorization: Bearer publicand allx-opencode-*headers were the same.Decisive test: in a local copy of the failing binary, patched only the version literal (
1.18.31.r12.g88c6c7a→1.18.31, same byte length, nothing else) → free tier works immediately. The version string is the discriminator.Implication: any development/nightly/
-gitbuild whose version is a git-describe string (<tag>.rN.gSHA) is not valid semver and gets rejected. The1.17.0 or newer is requiredvariant (#49430, #49431) looks like the same version-parsing gate with a different message.Suggested fixes (either side): report a sanitized semver in
User-Agentfor non-release builds (keep the full git-describe string in--version/logs), or make the server tolerate such strings / gate on an explicit client marker instead of the version.joaomj commentedon Sep 17, 2026
A third-party client (
pi) hits the same gate, and theUser-Agentversion is only one of two checksThanks @LouisLau-art for isolating the
User-Agentcheck. I hit the same error from a different client and reproduced it on purpose. This adds a second, independent condition the gate enforces.Context
piis a coding agent with a built-inopencodeprovider. It sendsx-opencode-client: pifor attribution. Since 2026-09-17 the free tier returns for every free model:The error appears on two hosts (macOS and Windows), with the same key and the same
piversion, and without the OpenCode CLI involved. That is why I looked at the request fingerprint rather than at the account.Method
I sent direct requests to
https://opencode.ai/zen/v1/responseswithAuthorization: Bearer public(anonymous free tier) and a minimal JSON body, then varied one header at a time.Finding 1 —
User-Agentmust beopencode/<recent semver>User-Agentopencode/1.18.31opencode/latest/1.18.31/cliopencode/999.0.0OpenCode/1.18.31opencode/0.0.0opencode/latestopencode/opencode(no slash)pi (darwin 25.6.0; arm64)curl/8.7.1OpenAI/JS 6.40.0This matches #49430/#49431 (
1.17.0 or newer is required) and the git-describe rejection. The trailing SDK tags (ai-sdk/provider-utils/... runtime/bun/...) are not required: plainopencode/1.18.31passes, andOpenCode/1.18.31passes, so the match is case-insensitive.Finding 2 —
x-opencode-sessionmust be an OpenCode session IDWith
User-Agent: opencode/1.18.31fixed, the session header is a second independent gate. The accepted format isses_+ 12 hex characters + 14 base62 characters, i.e. whatIdentifier.createproduces inpackages/opencode/src/id/id.ts.x-opencode-sessionses_<12 hex><14 base62>ses_+ 12 hex time +AAAAAAAAAAAAAAses_AAAAAAAAAAAA+ 14 random base62<uuid>(e.g.01a0af13-c6df-73c6-8803-9e6d9c720da7)ses_prefixNotes from the same tests:
Acomponent (which decodes to ~1971) fails. So the gate looks for a plausible timestamp, not an exact one.x-opencode-clientis not validated.cliandpiboth pass with a valid session.x-opencode-requestandx-opencode-projectare not required.{"model","input","stream"}payload behave the same.Bearer publicand a real account key both pass with a valid fingerprint.So a request to the free tier needs both an
opencode/-styleUser-Agentand ases_-formatted session ID. A third-party client that reports an honestUser-Agent(aspidoes) is rejected twice: once on the UA and once on its own UUID session.Request
Could the OpenCode team make the free-tier client requirements explicit in the docs? Concretely:
pialready sendsx-opencode-client: pi, which suggests third-party attribution was intended.User-Agentform and the requiredx-opencode-sessionformat in one place, so client authors can comply instead of reverse-engineering the edge service? A1.17.0 or newerminimum and a session-ID requirement would be enough.I am happy to test any documented header contract and to adjust
piaccordingly.feat: a lane that stops working while its vendor is silent gets a dat…
GooseOb commentedon Sep 17, 2026
@joaomj unlike opencode, pi works for me tho
joaomj commentedon Sep 17, 2026
@GooseOb which provider are you using in pi? The exact provider (opencode, opencode-go, opencode-zen etc).
GooseOb commentedon Sep 17, 2026
@joaomj oops, I tried
opencodeand acutally got the errorApparently I was using
opencode-free, which was provided by pi-free extensionTranberry commentedon Sep 17, 2026
Was working then suddenly got the warning, updated to latest version still get the warninf.
OS: Windows 11
Terminal: PowerShell 7.6.5
Simonko-912 commentedon Sep 17, 2026
Currently on newest ver, other agents running rn work with same model (big pickle free), but this single one at compaction got the same error
SuperYosh23 commentedon Sep 17, 2026
This is really pissing me off as i'm trying to work on a project, please fix this asap.
I am also using arch linux (omarchy)
FortuneDan commentedon Sep 17, 2026
same issue on big pickle, it specifically happens when it goes to perform compaction. Haven't changed anything in weeks, been using the same version and model and now suddenly today this happens
Simonko-912 commentedon Sep 17, 2026
For me, in the same session, i tried everything with big pickle, and nothing worked, still got the same error, and some other sessions simply work
michaelcoyote commentedon Sep 17, 2026
Yep, just started for me too after a few weeks of it working fine:
opencode version is 1.18.30
tty is Ghostty v1.3.1
OS is currently MacOS.
I'm guessing this isn't on my end though the way everyone is showing up here at once.
SuperYosh23 commentedon Sep 17, 2026
Yeah, this was working fine for me yesterday also but now it's not
38 remaining items
Corniel commentedon Sep 22, 2026
WIth this version I can not longer reproduce this issue.
gainfully4302 commentedon Sep 23, 2026
The Plan mode agent is working for me now but my read only Q/A agent that I created with markdown is triggering the "OpenCode's free tier can only be used..." error, so you might be on to something.
joshportman commentedon Sep 23, 2026
Thanks for investigating this. We can confirm a permission-dependent reproduction on Linux with the official npm CLI 1.18.32, model
opencode/big-pickle, and a configured Zen API credential.Failing command in an empty project:
This returns HTTP 403 /
FreeTierError: "OpenCode's free tier can only be used from within OpenCode". We also observed the same error on 1.18.31. Failed request log ID:bcc8d6a8-aa88-48f0-9152-2b9cb5f08e49.Using approval-gated tools instead of removing them resolves it here:
Result:
BIG_PICKLE_OK, exit 0. Confirmed again after persistingpermission: {"*":"ask", ...}in the global config and removing the environment override. Both successful requests reported cost 0, and neither executed a tool. No client/header spoofing, custom endpoint, or paid-model fallback was used.Could the free-tier validation support legitimate deny/reduced-toolset configurations, or at least return an actionable error explaining the incompatibility? The current message led us to investigate account authentication and client versions unnecessarily. Happy to provide additional sanitized details. Thank you!
orangeguitar commentedon Sep 23, 2026
Reproduction: free-tier gate fails when an agent sets
permission.bash: denyOpenCode: 1.18.32 (Linux)
Provider:
opencode(Zen free models:mimo-v2.6-flash-free,big-pickle)Not reproduced on:
opencode-gofree models (different header branch)Summary
Custom agents with
permission: bash: denyreliably get:The same agent with
bash: ask(or without abashpermission) succeedsimmediately. Built-in
plan/buildsucceed. This looks like the free-tierclient gate misclassifies requests when bash is fully denied — not a missing
User-Agent/session header from an external harness.
Controlled A/B/A/B (same binary, same model, ~2026-09-23)
Agent file
~/.config/opencode/agents/chat.md(custom,mode: primary):denyaskMinimal repro agent:
Change to
bash: ask→ succeeds. Pattern form also fails:Other single-key tests with
bashomitted all succeeded on Zen free:edit: deny,write: deny,lsp: deny,todowrite: deny,tavily_: ask,temperature: 0.5.Cross-agent: a throwaway custom agent with only
bash: denyalso failed; aminimal custom agent without a
bashpermission succeeded. So it is notspecific to the agent name
chat.Other observations
gpt-5.4-nano→Model access is disabled(separate issue).bash: ask(or omit bash) on agents that must use Zen free models.chat+ Zen free on this machine was 2026-09-08 (before the gate).Environment
opencode --version: 1.18.32chat.headershooks found)Happy to run more header-level tests if maintainers need them.
Eriol1805 commentedon Sep 23, 2026
Confirmed the workaround on Windows, Chocolatey 1.18.32: changing the agent's
permission.bashfromdenytoaskresolves the 403 for theplanagent withopencode/big-pickle(verified both via CLIopencode run --agent planreturning the expected reply at cost 0, and in interactive plan-mode sessions).edit/writeremaindeny, so the agent is still effectively read-only — every execution prompts for approval.Thanks to @orangeguitar for isolating the trigger to
bash: deny, and to @sekedus, @joshportman and @gainfully4302 for the reproductions. The current error message (OpenCode's free tier can only be used from within OpenCode) really sends you down the wrong path (account/auth/client version) — an actionable message about the toolset requirement would save a lot of debugging. Thanks!masuda-masuo commentedon Sep 24, 2026
Adding a precise bisect in case it helps, since you asked about custom setups.
Setup: opencode 1.18.31 (official npm release, Linux/WSL2). Logged into Zen + Go. opencode is driven headlessly (
opencode serve/opencode run) by an orchestrator. The worker agents are custom primary agents that start from"*": denyand only allow MCP tools that operate inside a sandbox container. Built-inbash/readare denied on purpose, so an automated worker can never touch the host.Repro: a minimal custom agent and
opencode run --agent <name> -m opencode/mimo-v2.6-flash-free "Reply with just OK", changing only one permission line each time:bash: denyread: denyedit/glob/grep/webfetch/task: deny"*": deny,bash: allow"*": deny,bash: ask,read: askSo the gate seems to require both
bashandreadto be present in the request's tool list.read: denyalone triggers it too, which I don't think has been reported here yet.askpasses because the tool stays in the list.The same agents keep working on paid models: 187 completed Go runs since 2026-09-19, 0 rejected. Every free model is rejected; before 2026-09-19 the same agents had 459 completed free runs.
Restricting bash/read is a safety choice for unattended runs, not a sign of a different harness. It would be great if denying those tools didn't count as "not OpenCode". Happy to provide more details if useful.
SamiNassim commentedon Oct 3, 2026
I just started getting this error too. I’m not sure what’s causing it, it started happening randomly.
Happy to provide more information needed to debug this.
macOS 26.6.2
OpenCode 1.18.34 (updated from 1.18.32 to see if that would fix it, but it didn’t)
Nothing custom.
I was using MiMo-V2.6-Flash. Tried to switch to Space Bunny but same thing.

juliancruzsanchez commentedon Oct 3, 2026
I am having this issue now
dl9sau commentedon Oct 3, 2026
Me too. "Error: Error from provider (Console): OpenCode's free tier can only be used from within OpenCode".
Had also this error some weeks ago. Resolved by upgrading opencode.
I just upgraded opencode on my macos (macports: version opencode @2.0.21 llm/opencode) -- previously 1.18.31 iirc. But this time the error stays.
raihanmd commentedon Oct 3, 2026
same here Error: Error from provider (Console): OpenCode's free tier can only be used from within OpenCode
❯ opencode --version
1.18.34
feat: steer users away from OpenCode free-tier models
thdxr commentedon Oct 11, 2026
Thanks for the report! OpenCode 2.0 is now out and development is focused on V2, so we're closing issues filed against OpenCode 1. Please upgrade (
opencode upgrade --major, details: https://anoma.ly/notes/opencode-2-0), and if this is still relevant on 2.0, reopen this issue or open a new one with your 2.0 version.— from 𝕺𝖕𝖊𝖓𝕮𝖔𝖉𝖊
k00shi commentedon Oct 17, 2026
Still happening in v2.
hossainchisty commentedon Oct 17, 2026
I'm using the official OpenCode Desktop app on macOS, but every prompt fails with this error:
Environment:
OS: macOS
OpenCode version: v2.0.32
Installed via: Homebrew
Troubleshooting tried:
Re-authenticated using opencode auth login opencode
Tested from /tmp to rule out project-specific configuration
Installed and tested the official Desktop app
Verified the installed version
The issue persists. Is this a known Console free-tier issue, and is there a recommended fix?