Morph

Introducing FlashCompact - the first specialized model for context compaction 33k tokens/sec 200k → 50k in ~1.5s Fast, high quality compaction

the general applied standard intelligence compute company

Join us in welcoming @morphllm Founder, @tejasybhakta to the AIE Miami lineup! Don't miss his talk 'Everything is Models' next week on the big stage! Get your tickets: ai.engineer/miami

warpgrep_github_search from @morphllm is probably the most unfathomoly unfair advantage you can have right now. 10x better than grep app Even beats Ctx7 tbh docs.morphllm.com/sdk/components…

@ryanleecode @kunchenguid @dhruvbhatia0 is working on a fix for the latest CC version, but we reccomend compacting at around 150k for best coding performance. for investigation, search, chat with repo type uses, no need to compact until 1m

@kunchenguid @morphllm What is your thoughts on the 1m context window question if we use the morph compact plugin?

for folks who feel their Claude Code got nerfed, here's what I've set in my ~/.claude/settings.json to make my CC's behavior more stable. snippet and explanation in thread below -

@tube_you86806 @FactoryAI oh wait actually no it doesnt. @FactoryAI wanna expose a compactions hook?

@tube_you86806 @FactoryAI yep! morphllm.com/setup

Introducing FlashCompact - the first specialized model for context compaction 33k tokens/sec 200k → 50k in ~1.5s Fast, high quality compaction

@steipete @Techguy0101010 @openclaw @steipete good to go? github.com/openclaw/openc…

@morphllm @Techguy0101010 @openclaw please change this to be a plugin

morphllm.com/mcp install the plugin + mcp here

Our Claude Code plugin is here! - WarpGrep for state of the art fast code search - FlashCompact, our specialized fast compaction model end to end speedup on long claude code sessions: -37%, while saving claude tokens and improving accuracy

@fastquant0 clerk outage with their bot protection feature. @clerk please fix

@morphllm hm any reason why it failed?

Agents don’t need bigger models. They need better tools. Morph trains coding subagents. Not for humans. For frontier models. Fast Apply edits at 10,000 tokens/sec. WarpGrep handles code and log search. Both keep the main model’s context clean Because when context gets too large, performance drops. Now Morph is pushing coding subagents even faster One newer model runs at 33,000 tokens/sec: docs.morphllm.com/sdk/components… 🎙️ @tejasybhakta, Founder & CEO, @morphllm on @fondocom @thestartpod w/ @davj

@ryanleecode opencode plugin github.com/morphllm/openc… working on claude code still. 1M limit for context window. should be able to support 5M in a week or so

@morphllm MCP support? and is there a context window limit?

@jackjoliet you can specify that in the sdk if you use the messages input docs.morphllm.com/sdk/components…

@morphllm keeps last N full messages?

We just removed access control. Anyone can now try this from the API as well. docs.morphllm.com/sdk/components…

@turbo_xo_ @sama 💚

@morphllm @sama i love codex but back when i used to use cursor WarpGrep earned a special place in my heart now i love Morph

@hari65535 yep! trained for both

@morphllm Only for code? Can this be used for summarising chat too?

Perfect compaction is a prerequisite for long-running agents. It’s the difference between a country of geniuses and a pile of clankers. #unLobotomizeClaude

@scent_fetish_ Not in the mcp yet - just npm for now. For opencode we have a plugin github.com/morphllm/openc…

Those guys are doing it right. @morphllm Do we update MCPs, or do they just work server-side (NPM use, not NPX)?

@rudzinskimaciej @itspers @albfresco @mots_pod Plugin coming soon

@morphllm @itspers @albfresco @mots_pod I would test it it would be great for pi or many claws