Here reasoning didn’t even finished. It’s 3rd time in this particular moment of the chat and 10th time in this chat alone (sometimes produces partial message, just like partial reasoning - seen here). Sometimes it shows Error.
Kimi K2.5 thinking. If I click retry and “same”, it will go to Kimi K2.5 Instant.
I don’t know if I should make it a new bug report.
in this particular case from the logs it seems like the model provider had an error mid stream. This is unfortunately possible, and more likely with models served from non-major labs. I have made a note of the provider that served this request, and will keep an eye out for continued failed responses to block the provider if needed.
Sapphire Crab
Apologies for late response - that was what I thought, since that day Kimi K2.5 (not that I use any other model) was doing it on average every few minutes. I didn’t know if you were hosting the model yourself, from 3rd party or from Moonshot themselves. I think that it was either on the same day or close to when I had the most spectacular failure and I had o stop the model, because it was looping. Since I looked away from the monitor, I only noticed it at like 8th full reply - which included activating web agent and doing searches. If I was on 8$ tier, it would actually be a big hit to my ability to do anything. Model after that could see it in it’s context. I had to abandon that thread, because context window was used significantly and pruning the thread would be a significant work. Obviously I don’t blame you. Such things happen. Now I have to find a way to make folders for the threads, because I have more pinned ones than old ones ;) I can’t even find the chat I was talking about.
When using T3 Chat with the reasoning model enabled, assistant replies sometimes show only the hidden “thinking” section and the UI displays “Message is empty” instead of the actual assistant message. This seems to happen when the internal reasoning payload is very long (e.g., for complex SQL analysis). If the assistant is asked to “keep thinking to a minimum” and answer briefly, messages appear normally again. Noticeable on models such as Model: GPT‑5.1 (Reasoning). Roughly what’s probably happening:
The model sends back:
A big hidden “reasoning” section.
A shorter “final message” that should be shown to you.
When the reasoning section is very long, the T3 Chat front‑end or API:
Either hits a size/serialization limit, or
Mis-parses the response so that the visible message field ends up null/empty.
The UI then displays “Message is empty”, even though the model did send an answer.
Copper Supernova
Copper Supernova
You can check the logs for this chat URL where the bug occurred:
This appears to be a case of hitting the output token limit while reasoning. Will look into rendering a better state for this.
Copper Supernova
Thanks for the update. One idea for improving the UX: Could T3 Chat enforce a hard token budget for the final, user‑visible answer?
For example, reserve a minimum number of X tokens for the answer itself.
Copper Supernova
Just to follow up on my earlier comment:
Would it also make sense to put an explicit cap on the reasoning budget itself? For example, given a total output limit, you could let the hidden reasoning use at most limit - min_answer_tokens, so there are always at least min_answer_tokens left for the user‑visible answer.
That way the model can’t burn the entire quota on internal reasoning, and in the worst case the user just gets a shorter, more high‑level answer instead of an empty message. Curious if that fits with how your reasoning vs. visible streams are wired up, or if there are constraints that make it tricky?
Mark Florkowski
Team
It would make a lot of sense for the API to be designed this way, but unfortunately most if not all are not
Salmon Jaguar
As of today (2026-02-18) I am getting this on almost every invocation of Sonnet 4.6, Sonnet 4.6 (Thinking), and Opus 4.6, regardless of the complexity of the request, AFAICT.
Mark Florkowski
Team
Can you provide the urls of the chats where you encoutnered this so that I can checl logs?
Salmon Jaguar
So this is the one that made me hit up the forum: https://t3.chat/chat/2f3362df-074e-4290-b35e-47e4fca312b4 BUT: I did reruns, so I don’t know if you can see half the empty AI messages. I assume you can see the chat w/o me giving perms?
Mark Florkowski
Team
I can’t see the chat itself, but I can see logs. Will look into this
Aqua Volcano
Still happening with 3.1 pro
Mark Florkowski
Team
That model in particular is having a lot of issues. We added a warning about this, as the issue is on Google’s side and we are doing what we can to mitigate it. I highly recommend not using search with this model for now.
Log in to comment and vote
Comments36
Blue Brownie
again the same thing
dude this consumes premium requests having to retry too does only to get empty response.
Mark Florkowski
Can you please provide URLs of chats where you are encountering this issue so that I can check the logs?
Crimson Prairie
It’s happened 3 out of 3 times today using Gemini 3 Pro, other models doesn’t have this problem
Mark Florkowski
Can you provide an example url of a chat where this issue occurred for you?
Crimson Prairie
Yup. Not sure why I only found two, I swear I counted three, oh well
https://t3.chat/share/amcz2tssf7
https://t3.chat/share/lo0qzk98nm
Copper Supernova
https://t3.chat/chat/ffa36b2b-f08f-47b6-b160-483c62f54ca2
Teal Astronaut
https://t3.chat/share/972gz65trq
Same issue here, with Gemini 3 Pro on High
Sapphire Pinwheel
The issue persist with Gemini 3 model
Pink Darkmatter
I experience this issue frequently when using Gemini 3 Pro.
Chat references:
https://t3.chat/chat/c946b333-ec67-4587-a9f8-b08179b46ba0 (
https://t3.chat/share/2hm3wcdecg)
https://t3.chat/chat/ab3c5e92-b083-4691-a735-e11a3624b2f6 (
https://t3.chat/share/lx3lfhw99p)
Sapphire Crab
Here reasoning didn’t even finished. It’s 3rd time in this particular moment of the chat and 10th time in this chat alone (sometimes produces partial message, just like partial reasoning - seen here). Sometimes it shows Error.
Kimi K2.5 thinking. If I click retry and “same”, it will go to Kimi K2.5 Instant.
I don’t know if I should make it a new bug report.
Link:
https://t3.chat/chat/8b2db4ba-2fed-4365-9d11-f0e9888a9498
Mark Florkowski
in this particular case from the logs it seems like the model provider had an error mid stream. This is unfortunately possible, and more likely with models served from non-major labs. I have made a note of the provider that served this request, and will keep an eye out for continued failed responses to block the provider if needed.
Sapphire Crab
Apologies for late response - that was what I thought, since that day Kimi K2.5 (not that I use any other model) was doing it on average every few minutes. I didn’t know if you were hosting the model yourself, from 3rd party or from Moonshot themselves. I think that it was either on the same day or close to when I had the most spectacular failure and I had o stop the model, because it was looping. Since I looked away from the monitor, I only noticed it at like 8th full reply - which included activating web agent and doing searches. If I was on 8$ tier, it would actually be a big hit to my ability to do anything. Model after that could see it in it’s context. I had to abandon that thread, because context window was used significantly and pruning the thread would be a significant work. Obviously I don’t blame you. Such things happen. Now I have to find a way to make folders for the threads, because I have more pinned ones than old ones ;) I can’t even find the chat I was talking about.
Amber Fuelcell
Merged request
Message is Empty when reasoning is long
When using T3 Chat with the reasoning model enabled, assistant replies sometimes show only the hidden “thinking” section and the UI displays “Message is empty” instead of the actual assistant message. This seems to happen when the internal reasoning payload is very long (e.g., for complex SQL analysis). If the assistant is asked to “keep thinking to a minimum” and answer briefly, messages appear normally again. Noticeable on models such as Model: GPT‑5.1 (Reasoning). Roughly what’s probably happening:
The model sends back:
A big hidden “reasoning” section.
A shorter “final message” that should be shown to you.
When the reasoning section is very long, the T3 Chat front‑end or API:
Either hits a size/serialization limit, or
Mis-parses the response so that the visible message field ends up null/empty.
The UI then displays “Message is empty”, even though the model did send an answer.
Copper Supernova
Copper Supernova
You can check the logs for this chat URL where the bug occurred:
https://t3.chat/chat/ffa36b2b-f08f-47b6-b160-483c62f54ca2
Mark Florkowski
This appears to be a case of hitting the output token limit while reasoning. Will look into rendering a better state for this.
Copper Supernova
Thanks for the update. One idea for improving the UX: Could T3 Chat enforce a hard token budget for the final, user‑visible answer?
For example, reserve a minimum number of X tokens for the answer itself.
Copper Supernova
Just to follow up on my earlier comment:
Would it also make sense to put an explicit cap on the reasoning budget itself? For example, given a total output limit, you could let the hidden reasoning use at most
limit - min_answer_tokens, so there are always at leastmin_answer_tokensleft for the user‑visible answer.That way the model can’t burn the entire quota on internal reasoning, and in the worst case the user just gets a shorter, more high‑level answer instead of an empty message.
Curious if that fits with how your reasoning vs. visible streams are wired up, or if there are constraints that make it tricky?
Mark Florkowski
It would make a lot of sense for the API to be designed this way, but unfortunately most if not all are not
Salmon Jaguar
As of today (2026-02-18) I am getting this on almost every invocation of Sonnet 4.6, Sonnet 4.6 (Thinking), and Opus 4.6, regardless of the complexity of the request, AFAICT.
Mark Florkowski
Can you provide the urls of the chats where you encoutnered this so that I can checl logs?
Salmon Jaguar
So this is the one that made me hit up the forum:
https://t3.chat/chat/2f3362df-074e-4290-b35e-47e4fca312b4
BUT: I did reruns, so I don’t know if you can see half the empty AI messages. I assume you can see the chat w/o me giving perms?
Mark Florkowski
I can’t see the chat itself, but I can see logs. Will look into this
Aqua Volcano
Still happening with 3.1 pro
Mark Florkowski
That model in particular is having a lot of issues. We added a warning about this, as the issue is on Google’s side and we are doing what we can to mitigate it. I highly recommend not using search with this model for now.