Gemini 3 seems to have a much smaller token output limit than 2.5. I used to use Gemini to restructure essays into an LLM-style format to improve readability, but the Gemini 3 release was a huge step back for that particular use case.

Even when the model is explicitly instructed to pause due to insufficient tokens rather than generating an incomplete response, it still truncates the source text too aggressively, losing vital context and meaning in the restructuring process.

I hope the 3.1 release includes a much larger output limit.

▲

esafak 5 hours ago | parent | next [-]

People did find Gemini very talkative so it might be a response to that.

▲

NoahZuniga 4 hours ago | parent | prev | next [-]

Output limit has consistently been 64k tokens (including 2.5 pro).

▲

jayd16 5 hours ago | parent | prev | next [-]

> Even when the model is explicitly instructed to pause due to insufficient tokens

Is there actually a chance it has the introspection to do anything with this request?

▲

maxloh 3 hours ago | parent | next [-]

Yeah, it does. It was possible with 2.5 Flash.

Here's a similar result with Qwen Qwen3.5-397B-A17B: https://chat.qwen.ai/s/530becb7-e16b-41ee-8621-af83994599ce?...

	▲	jayd16 3 hours ago \| parent [-]
		Ok it prints some stuff at the end but does it actually count the output tokens? That part was already built in somehow? Is it just retrying until it has enough space to add the footer?

▲

verdverm 3 hours ago | parent | prev | next [-]

No, the model doesn't have purview into this afaik

I'm not even sure what "pausing" means in this context and why it would help when there are insufficient tokens. They should just stop when you reach the limit, default or manually specified, but it's typically a cutoff.

You can see what happens by setting output token limit much lower

▲

otabdeveloper4 3 hours ago | parent | prev [-]

No.

▲

MallocVoidstar 4 hours ago | parent | prev [-]

> Even when the model is explicitly instructed to pause due to insufficient tokens rather than generating an incomplete response

AI models can't do this. At least not with just an instruction, maybe if you're writing some kind of custom 'agentic' setup.

	▲	maxloh 3 hours ago \| parent [-]
		Yeah, it does. It was possible with 2.5 Flash. Here's a similar result with Qwen Qwen3.5-397B-A17B: https://chat.qwen.ai/s/530becb7-e16b-41ee-8621-af83994599ce?...