ManjuAi
Run deepresearch, coding agents for free.
9 followers
Run deepresearch, coding agents for free.
9 followers
Run deepresearch, coding agents for free using Manju. User have control over the length of the report deepresearch agents provides. Models have complete context of when using the coding agents. Get almost all premium models for just 5.3 dollars.
Free
Launch Team / Built With
Viktor.comAn AI coworker that actually does the work
Promoted
How do you keep the costs down enough to offer premium models for $5.3, and which specific models are included at that price tier?
@ayebykbozkqseo You get access to almost all of gpt family models like gpt 5.4, gpt 5.3 codex, chat and so on. You get access to all of gemini models, anthropic 4.6 sonnet, haiku, anthropic 4.5. Thank you for pointing that out we will update the model selector to make sure users can see what all models they access.
How does the coding agent handle context across really long sessions, like multi-day projects where the model would normally lose track of earlier decisions?
@mit1626164 All the files of the projects are stored into a vector database when user adds a file it again gets embedded into the vector database. And when the user queries something related to the project, the agent pulls all the required information from the database and responds to the user. As you can see this project was built by openai/gpt-oss-120b:free, which is a free model available in open router. Its a pretty big project and the agent never losses context.
The deep research tool actually lets you tweak the report length which is super useful when you just need a quick summary. Coding agents keeping full context across sessions is a nice touch too.
@beyzatav7 Glad to hear that those had been one of our main focus while building.
Tried the coding agent today and it kept full project context without me re-explaining anything, which is rare for free tools. Pricing under six bucks for most premium models feels too good to be true but it actually works.
@ftavukcuyo70638 Thank you for the review, we are trying to help users get access to great tools for as a less cost. This is possible since all models need is context, when given they respond really well.