Announcing: A Free Stealth Model from a Top AI Model Lab
A new stealth model from one of the top 10 AI model labs in the world.
Today we’re announcing a new stealth model from one of the top 10 AI model labs in the world.
Kilo Code is the first AI coding agent in the world to have early access to this model.
The free, stealth model is called Spectre and you can easily find it in Kilo Code’s model selector:
Model details
This model is optimized for software development tasks:
Context window: 256,000 tokens
Maximum output: 8,000 tokens per step
Currently no image input or multimodal support, but expect updates soon
Our Impressions So Far
We’ve had positive experiences testing the model. We successfully:
Built an impressive 2D game in one shot
Debugged a hard-to-find memory leak inside a Rails application
Migrated from an old NextJS 12 project
This might be your new favorite AI model
The team behind this model is extremely ambitious and you can clearly see that once you start using the model.
For a start, there are no usage caps when you use the model. Yes, this means you get (literally) UNLIMITED access to the model, so make it burn!
Once we reach our first 100 million tokens with Spectre, we’ll reward $500 in Kilo Code credits to 3 people who submit this form and show us something cool they’ve built with Spectre.
We’ll announce the winners and share our first impressions of this model on tomorrow’s free webinar. See you there!




It's not free, and I've already been charged money when using it!
Spectre was a total disappointment for us. We have a series of prompts that we use for testing each LLM and this model failed miserably. Horrible at tool calling. Hallucinating more than any model we have every worked with. Very hard to steer. Something is wrong with this model, it needs to be fixed. I don't understand the hype, because it fails at any real world challenges.