Models & Labs
Inception launches Mercury 3 diffusion model
The Rundown AIhigh confidence
Why it matters
- →Diffusion models offer a viable alternative to autoregressive architectures for high-speed text generation.
- →Mercury 3 demonstrates that non-traditional LLMs can match frontier performance metrics at lower latency.
- →The launch signals intensifying competition in the inference speed and cost optimization market.
AI startup Inception has launched Mercury 3, a diffusion-based language model designed to compete with leading autoregressive systems like OpenAI's GPT-6 Luna and Anthropic's Claude Haiku 5.5. Unlike traditional models that generate text sequentially, Mercury 3 uses a diffusion process to draft multiple tokens simultaneously, achieving speeds of over 1,000 tokens per second while maintaining competitive accuracy on standard benchmarks. The model is priced at $0.20/$0.70 per million input/output tokens, positioning it as a cost-effective option for high-throughput applications such as live tutoring and voice support agents. Co-founder Stefano Ermon stated that the company views compute and data as the primary scaling limits rather than algorithmic barriers, with a successor model already in training.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
More in Models & Labs
Models & Labsmodels
llama.cpp b12029 adds CUDA 13 and Snapdragon support
This release quietly cements llama.cpp as the universal inference runtime by finally supporting NVIDIA's latest CUDA 13 stack alongside AMD's ROCm 10.0. The inclusion of native builds for Linux arm64 with Snapdragon chips, covering CPU, Adreno GPU, and Hexagon NPU, signals a serious push into mobile edge deployment. While Apple Silicon KleidiAI is disabled in this specific build, the broader hardware coverage means developers no longer need to compile from source for most modern accelerators. It’s less about new features and more about removing friction for anyone running local models on bleeding-edge or diverse hardware.
llama.cpp Releases
© The Verge AIModels & Labsother
