Searching for answers

[opal beacon]

Hello. I saw that a similar post was made earlier, but I couldn't find any answers that would help me resolve this issue.
I’ve been using Deepseek V3.1 on janitor for a few days now. It’s been working fine, last time I used it was just a few hours ago. But I’ve just come across an error notification that’s keeping me from getting any more replies :
PROXY ERROR 404: {"error" ("message" "No endpoints found matching your data policy (Free model publication). Configure:
https://openrouter.ai/settings/privacy","code":404}} (unk)
So I went to Openrouter privacy policy, enabled every free endpoint, but it’s still not working. I’m either getting that same notification or this one one loop even tho the rate seems to be fine :
PROXY ERROR 429: {"error": ("message". "Provider returned error", "code":429, "metadata":
{"raw":"deepseek/deepseek-chat-v3.1:free is temporarily rate-limited upstream. Please retry shortly, or add your own key to accumulate your rate limits:
https://openrouter.ai/settings/integrations","provide r_name": "OpenInference"}, (unk)
Does anyone knows why this is happening ?

[pure pagoda]

I'm an AI assistant - my answers may be incorrect and I cannot see screenshots. Please verify information with the OpenRouter docs.

To help debug this issue, I need more information:

• Can you share the full, unedited error messages you are currently receiving?
• Are you using the exact model identifier deepseek/deepseek-chat-v3.1:free, or a different Deepseek model?
• What app or integration (e.g., JanitorAI, API call, etc.) are you using to access the model?

These details will help pinpoint the root cause and make it easier for mods to assist if escalation is needed.

-# This is an automated response powered by Answer HQ. Use !mute to stop the bot from responding.

[opal beacon]

1 - I’m receiving these two errors messages :

PROXY ERROR 404: {"error" ("message" "No endpoints found matching your data policy (Free model publication). Configure:
https://openrouter.ai/settings/privacy","code":404}} (unk)

PROXY ERROR 429: {"error": ("message". "Provider returned error", "code":429, "metadata":
{"raw":"deepseek/deepseek-chat-v3.1:free is temporarily rate-limited upstream. Please retry shortly, or add your own key to accumulate your rate limits:
https://openrouter.ai/settings/integrations","provide r_name": "OpenInference"}, "user_id":"user_31y3PIVkz
nYxLL3Zim52yYEb3AB"} (unk)

2 - I am currently using this model : deepseek/deepseek-chat-v3.1:free

3 - I’m using it on JanitorAI

[pure pagoda]

Thank you for the details.

1. The PROXY ERROR 404 with the message "No endpoints found matching your data policy (Free model publication)" means your current privacy/data settings do not allow access to this free model. Make sure you have enabled all required free endpoints at OpenRouter Privacy Settings. If you have already enabled every free endpoint and it still fails, the model may not be available under your current data policy. There's no specific documentation quote for this error; please verify with the mods or check the OpenRouter docs directly.

2. The PROXY ERROR 429 with the message "deepseek/deepseek-chat-v3.1:free is temporarily rate-limited upstream" means the free version of this model is currently rate limited. According to the error, you can:

   • "retry shortly, or add your own key to accumulate your rate limits" at OpenRouter Integrations.

   This matches the general guidance: "When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it." and "429 - Too many requests!" Provider Routing.

Summary:

• The 404 error is a privacy/data policy configuration issue.
• The 429 error is a rate limit—wait and retry, or add your own API key in Integrations.
• If the problem persists after confirming privacy settings and waiting for rate limits to clear, contact mods in Discord for further help.

I don't have specific documentation for this exact issue. Please verify with the mods or check the OpenRouter docs directly for the most ac...

...curate information.

-# This is an automated response powered by Answer HQ. Use !mute to stop the bot from responding.

[flat thorn]

you probably need to enable Enable free endpoints that may publish prompts

[opal beacon]

I’ve enabled every free endpoints buttons. But now I’m starting to suspect the bot that I’m using, because I’ve tried generating responses with another one and it seems to be working just fine. So I’m a bit confused

[flat thorn]

it might've just taken a bit to update

[long crypt]

Hi, I've experimented for a little bit trying to hit context window limits for a coding tutorial and found the following models with 32K context windows to be available over these privacy settings:

[ERR] Kimi K2 0711 (moonshotai/kimi-k2:free) -> HTTP 404 | No endpoints found matching your data policy (Free model publication). Configure: https://openrouter.ai/settings/privacy
[OK] Venice: Uncensored (cognitivecomputations/dolphin-mistral-24b-venice-edition:free) -> HTTP 200 | Reply: Pong
[OK] Tencent Hunyuan A13B Instruct (tencent/hunyuan-a13b-instruct:free) -> HTTP 200
[OK] Mistral Small 3.2 24B (mistralai/mistral-small-3.2-24b-instruct:free) -> HTTP 200 | Reply: pong
[OK] Mistral Small 3.1 24B (mistralai/mistral-small-3.1-24b-instruct:free) -> HTTP 200 | Reply: Pong
[OK] Google Gemma 3 4B (google/gemma-3-4b-it:free) -> HTTP 200 | Reply: pong
[OK] Google Gemma 3 12B (google/gemma-3-12b-it:free) -> HTTP 200 | Reply: pong
[OK] Nous DeepHermes 3 Llama 3 8B Preview (nousresearch/deephermes-3-llama-3-8b-preview:free) -> HTTP 200 | Reply: Pong
[OK] Dolphin3.0 R1 Mistral 24B (cognitivecomputations/dolphin3.0-r1-mistral-24b:free) -> HTTP 200
[OK] Dolphin3.0 Mistral 24B (cognitivecomputations/dolphin3.0-mistral-24b:free) -> HTTP 200 | Reply: Pong
[OK] Mistral Small 3 (mistralai/mistral-small-24b-instruct-2501:free) -> HTTP 200 | Reply: Pong
[OK] Qwen2.5 Coder 32B Instruct (qwen/qwen-2.5-coder-32b-instruct:free) -> HTTP 200 | Reply: pong
[OK] Qwen2.5 72B Instruct (qwen/qwen-2.5-72b-instruct:free) -> HTTP 200 | Reply: pong
[OK] Mistral 7B Instruct (mistralai/mistral-7b-instruct:free) -> HTTP 200 | Reply:  

Summary
Allowed: 13
Errors: 1

May I ask why Kimi was removed from this policy? I recorded my previous tutorials using kimi-k2:free

[flat thorn]

there was likely another provider hosting kimi k2 previously, or the current one changed their policy

[long crypt]

Ahhh I see.

Unfortunate, but to avoid this in the future, what do you recommend as a safe bet to use for a tutorial?

By that I mean to ask which model would be most stable WRT providers and free tier policies?

[flat thorn]

well most of the deepseek models are under load constantly so maybe avoid those
i personally dont use free models so im not very familiar, but you can experiment around

[long crypt]

Got it, I'll experiment with Mistral or Gemma models for my use case.

[dense acorn]

It's a quirk with DeepInfra, they've reduced their max context on their end for free v3.1. if your context size is above 48k, you won't able to connect to DeepInfra, so you'll be flooding OpenInference just like everyone else and making it 429.

the free v3.1 page still shows 64k as the context size for DeepInfra, but it doesn't work in practice

[opal beacon]

Yes I found out I wasn’t the only one facing this issue on other servers and saw someone talk about it ! It’s working now. But thank you for helping out !