Continue.Dev - Revision history

1G-N15 at 14:33, 19 January 2026

2026-01-19T14:33:54Z

← Older revision		Revision as of 14:33, 19 January 2026
Line 1:		Line 1:
	== Older Versions are Better ==		== Older Versions are Better ==

	Not to harp on the devs too much, it is a FOSS project after all, but the old versions (pre-1.0) are just so, so much better than the new ones, if you don't care about agentic AI too much (and honestly, for those, just use [[Roo Code]]). It's much stabler, autocomplete is much faster and more reliant, and in general it just works well with local AI for the GPU-poor (e.g. by using indexing instead of agentic, which local smaller LLMs might still struggle with).		Not to harp on the devs too much, it is a FOSS project after all (and thus can easily be forked), but the old versions (pre-1.0) are just so, so much better than the new ones, if you don't care about agentic AI too much (and honestly, for those, just use [[Roo Code]]). It's much stabler, autocomplete is much faster and more reliant, and in general it just works well with local AI for the GPU-poor (e.g. by using indexing instead of agentic, which local smaller LLMs might still struggle with).

			TL;DR: Older versions of Continue.Dev are less buggy for autocomplete/chat, while newer versions focus on agentic programming, which I don't use (or will just use [[Roo Code]] if I need to). So I don't really see a point in using the newer versions of Continue.Dev.

1G-N15 at 14:32, 19 January 2026

2026-01-19T14:32:41Z

← Older revision		Revision as of 14:32, 19 January 2026
Line 1:		Line 1:
	== Older Versions are Better ==		== Older Versions are Better ==

	Not to harp on the devs too much, it is a FOSS project after all, but the old versions (pre-1.0) are just so, so much better than the new ones, if you don't care about agentic AI too much. It's much stabler, autocomplete is much faster and more reliant, and in general it just works well with local AI for the GPU-poor (e.g. by using indexing instead of agentic, which local smaller LLMs might still struggle with).		Not to harp on the devs too much, it is a FOSS project after all, but the old versions (pre-1.0) are just so, so much better than the new ones, if you don't care about agentic AI too much (and honestly, for those, just use [[Roo Code]]). It's much stabler, autocomplete is much faster and more reliant, and in general it just works well with local AI for the GPU-poor (e.g. by using indexing instead of agentic, which local smaller LLMs might still struggle with).

1G-N15: Created page with "== Older Versions are Better == Not to harp on the devs too much, it is a FOSS project after all, but the old versions (pre-1.0) are just so, so much better than the new ones, if you don't care about agentic AI too much. It's much stabler, autocomplete is much faster and more reliant, and in general it just works well with local AI for the GPU-poor (e.g. by using indexing instead of agentic, which local smaller LLMs might still struggle with)."

2026-01-19T14:31:54Z

Created page with "== Older Versions are Better == Not to harp on the devs too much, it is a FOSS project after all, but the old versions (pre-1.0) are just so, so much better than the new ones, if you don't care about agentic AI too much. It's much stabler, autocomplete is much faster and more reliant, and in general it just works well with local AI for the GPU-poor (e.g. by using indexing instead of agentic, which local smaller LLMs might still struggle with)."

New page

== Older Versions are Better ==

Not to harp on the devs too much, it is a FOSS project after all, but the old versions (pre-1.0) are just so, so much better than the new ones, if you don't care about agentic AI too much. It's much stabler, autocomplete is much faster and more reliant, and in general it just works well with local AI for the GPU-poor (e.g. by using indexing instead of agentic, which local smaller LLMs might still struggle with).