Tenstorrent is where AI runs. Video, speech, code. Big or small.

Read the full release: tenstorrent.com/newsroom/ten…

Thank you Tokyo! 🇯🇵 Here’s everything we announced at TT-Deploy Japan: Faster AI Inference • Kimi K2.6 900 t/s/u, 3x faster than GPUs • LTX 2.3 Fast 6 sec video gen in ~6 sec, 144 frames, 1080p, 4x faster than GPUs • DeepSeek-R1-0528 671B 400 t/s/u TT-Ascalon S Available Today • A licensable RISC-V CPU built for the next generation of agentic AI applications Heterogenous or Stand Alone • Easily deploy Tenstorrent Galaxy alongside existing infrastructure or standalone • @aiand_'s sovereign heterogenous inference platform with Tenstorrent Galaxy™ superclusters

Run anywhere. SF, LA, TX, NY...Tokyo

Tenstorrent Galaxy™ Blackhole is designed to deliver flexible and scalable AI infrastructure with industry-leading performance across real-world AI workloads. Performance starts with data movement: - Unified compute, memory, and networking - Seamlessly data flow across chips and servers alike - No proprietary interconnects or rigid workload constraints The result is near-linear scaling. Adding Galaxies not only increases throughput, but lowers latency, enabling faster time-to-first-token at scale. Run AI inference at scale: tenstorrent.com/hardware/gal…

Introducing tt-vscode-toolkit, an interactive learning environment designed to accelerate developing on Tenstorrent hardware. The toolkit brings project templates directly into VS Code, with lessons covering model deployment, agent frameworks, video generation, and more. Get started: docs.tenstorrent.com/tt-vsco…

💡New Analyst Insight: AI inference is not shaping up to be a one-chip market. VP & Principal Analyst @MattKimball_MIS examines how @tenstorrent's TT-Deploy launch positions Galaxy Blackhole, RISC-V, open-source software, and cost-per-token economics for a more heterogeneous AI infrastructure era. Read more: moorinsightsstrategy.com/ana…

Run @tenstorrent kernels on your CPU. ttsim is a bit-exact, full-system simulator of Wormhole and Blackhole. The whole tt-metal stack on Linux, Mac (UTM/QEMU) or Windows (WSL2) — numerics match silicon bit-for-bit. Bring up kernels of explore the architecture. No silicon required. First of its kind. Now Open Source github.com/tenstorrent/ttsim

Very big TT-Boltz update! You can now connect multiple @tenstorrent machines together and TT-Boltz makes it all look like a single big machine. They just have to be in the same network. No expensive cables. You can connect anything. The video shows a cluster of QuietBox 1, QuietBox 2, and a workstation with a single Blackhole card, predicting protein structures together. This cluster of workstations has much higher throughput than an Nvidia B200, at less than half the cost. Of course, you can also combine multiple Tenstorrent Galaxy servers into a single cluster. They just have to be in the same network. Every day, a drug designed on Tenstorrent hardware becomes more likely. Incredible things to be announced soon.

4x cost reduction in TTS inference with @tenstorrent! 11 NVIDIA L40S ran 550 simultaneous audio-stream at ~$100K. Now, 27 Tenstorrent P100 chips do the same at ~$27k. First production-grade TTS to match the cost of text tokens without degradation in audio quality. Hear it straight from the team that built it: @AkshatMandloi10 and @ranjith_m_s in the video below.