Launch of our TIM-Qwen3.6-27B Model today. We took a great, small Qwen model and substantially improved it's performance:
- 1M+ context window (10x extension)
- Run 3x as many concurrent runs
- 75% less memory required
- 1.5x token throughput
- OpenAI / Anthropic SDK compatible
If you haven't tried the Subconscious playground, here's a look. I gave the model search tools, asked it to tell me about our research paper, and boom, we have a personalized deep research agent.
Took 15 seconds to make.
Today we're launching Subconscious: a new platform for building agents with long-horizon reasoning and tool use, backed by MIT research.
One API call. Tool use. Context beyond existing limits.
If you're building agents, let's talk.