Google TPU IronwoodGoogle
12
Inference-first TPU pods
- Version
- Ironwood (7th generation)
- Cost
- from $5.4/unit
- Model
- Gemini
Tenstorrent Blackhole
Tenstorrent · #1 most active of 18 in Chips & Compute
Open-source software stack and RISC-V based accelerators sold as cards and IP.
83
Trust score · Mostly verified
3 data points checked · 2026-09-14
Current version
Blackhole
Entry cost
from $999/unit
Changes / 30d
16
The short answers · verified September 15, 2026
- What is the latest version of Tenstorrent Blackhole?
- The current shipped version of Tenstorrent Blackhole is Blackhole, as of September 15, 2026.
- How much does Tenstorrent Blackhole cost?
- Tenstorrent Blackhole starts at $999 per month on its cheapest paid tier, as of September 15, 2026.
- What AI model does Tenstorrent Blackhole use?
- Tenstorrent Blackhole runs primarily on Tensix cores.
Source: Tenstorrent — github.com · Reusable under CC BY 4.0 — cite Tomorrow
Capabilities
- RISC-V cores
- open tt-metal stack
- Ethernet-scaled cards
- IP licensing
- Passive, active and liquid cooling options
- High-performance AI processing
- Liquid-cooled workstations (TT-QuietBox)
- Scale-out servers (Tenstorrent Galaxy)
- PCIe Cards
- Tensix Cores
- Liquid cooling options
- TT-Metalium low-level C++ SDK
- TT-NN high-level Python API
- TT-Forge MLIR-based compiler
- TT-QuietBox workstation
- Tenstorrent Galaxy server
- PCIe Card form factor
- Active, passive, and liquid cooling options
- Infinitely scalable
- TT-Metalium low-level C++ SDK support
- TT-NN high-level Python API support
- TT-Forge MLIR-based compiler support
- Liquid-cooled workstation options
- Low-level kernels (LLK)
- C++ SDK (TT-Metalium)
- Python API (TT-NN)
- MLIR compiler (TT-Forge)
- Cloud-Native Support (Kubernetes)
- Virtual Studio Code Toolkit (TT-VSCode-Toolkit)
- AI processing
- liquid cooled
- PCIe Card
- Tensix cores
- Liquid-cooled workstation option
- Liquid Cooled Workstations
- AI Processing
- liquid-cooled workstation support
- PCIe card support
- enterprise-class AI server support
- Liquid-cooled workstation
- Scalable AI server
- Workstations
- Liquid-cooled
- Low-level C++ SDK (TT-Metalium)
- MLIR-based compiler (TT-Forge)
- Ethernet routing configuration
- AI model serving
- AI acceleration
- liquid cooled workstations
- enterprise-class AI server
- open-source compiler support
- AI Acceleration
- Liquid Cooled Workstation Support
- PCIe Interface
- RISC-V Architecture
- Open-Source Software Stack
- RISC-V architecture
- Liquid Cooling Option
- AI hardware accelerator
- Liquid-cooled options
- PCIe Card compatibility
- Tensor Core / Tensix cores
- AI accelerator
- PCIe card
- Workstation AI
- Enterprise-class AI server
- liquid-cooled workstation
- LLM inference
- AI hardware
- Workstation
- Server
- Liquid-cooled Workstations
- Open-source software stack
- PCIe cards
- TT-Metalium SDK
- multi-server configurations
- liquid cooled workstation
- low-level kernels
- low-level C++ SDK
- high-level Python API
- AI Hardware
- Liquid Cooled Workstation
- AI Accelerator
- Liquid-cooled workstation support
- PCIe Card support
- PCIe card form factor
- Enterprise-class AI server options
- Workstation Integration
- liquid cooling option
- PCIe AI Processor
- C++ SDK / Metalium
- PyTorch/JAX/ONNX Support
- vLLM serving
- TTNN APIs
- Enterprise-class AI servers
- C++ SDK
- Python API
- MLIR Compiler
- MLIR Compiler (TT-Forge)
- Liquid-Cooled Workstations
- Enterprise AI Servers
- Servers
- Liquid Cooled
- Enterprise-Class AI Server
- Liquid-cooled option
- High-level Python API (TT-NN)
Pricing
Cards
Starting price for individual Blackhole cards
$999
Workstations (TT-QuietBox)
Whisper-quiet, liquid cooled deskside workstation
$9999
Servers (Galaxy)
Sovereign, scale-out servers for production AI
$70000
Entry price over time
Geek mode
Models under the hood
- inferred
Gemma-2
Google
- inferred
Gemma4
Google
- inferred
GLM-5.2
Zhipu AI
- inferred
Kimi-K3
Moonshot AI
- inferred
Llama-3.3-70B
Meta
- inferred
Qwen3-32B
Alibaba
- disclosed
Tensix cores
Tenstorrent
- disclosed
tt-metal / tt-nn
Tenstorrent
Context window
—
Public API
no
Multi-model routing
no
Stack signals
- C++
- Python
- PyTorch
- JAX
- ONNX
Shared model stack
Other tracked products running on the same foundation models — a quick read on how much of the catalog moves when one of these models changes.
- GLM-5.2+1 aliasZhipu AI · 5 products
- Kimi-K3+1 aliasMoonshot AI · 4 products
- Gemma-2Google · 1 product
- Gemma4Google · 1 product
- Llama-3.3-70BMeta · 1 product
- Qwen3-32BAlibaba · 1 product
- Tensix coresTenstorrent · 1 product
- tt-metal / tt-nnTenstorrent · 1 product
green disclosed · amber inferred · grey unattributed. Aliases are folded into one model; provider concentration counts only models with an established vendor.
Reported scale
Reference data. Each figure is whatever the source actually said — weekly users, downloads, revenue run-rate — with its own definition and date. These are not comparable between tools and are never used to rank anything.
No public usage figure on record for this product.
Also in Chips & Compute
All Chips & ComputeCerebras WSE-3
Cerebras
9
Wafer-scale inference and training
- Version
- WSE-3
- Cost
- Free tier / developer tier pricing available (contact/partner integrations)
- Model
- Codex-Spark
NVIDIA Blackwell
NVIDIA
9
GB200/B200 rack-scale AI systems
- Version
- Blackwell
- Cost
- Pricing is not publicly listed (enterprise hardware architecture/system)
- Model
- Blackwell
5
Cloud-native training silicon
- Version
- Trainium2
- Cost
- Pricing not published; available via AWS EC2 instance pricing.
- Model
- Neuron compiler
Intel Gaudi 3
Intel
5
Cost-focused training and inference accelerator.
- Version
- 1.24.1
- Cost
- Pricing not published. Available via Intel Tiber AI Cloud or OEM platforms.
- Model
- Accelerator
SambaNova SN40L
SambaNova
5
Dataflow chips for model serving
- Version
- SN40L
- Cost
- Free tier available. Paid plans include Developer (Pay-as-you-go) and Enterprise (Subscription-based).
- Model
- DeepSeek-V3.1
Change history
- version
- capability
New capabilities: Liquid-cooled option, High-level Python API (TT-NN)
104 tracked106 tracked · +Liquid-cooled option, High-level Python API (TT-NN)
source - version
- capability
New capabilities: Enterprise-Class AI Server
103 tracked104 tracked · +Enterprise-Class AI Server
source - version
- capability
- capability
New capabilities: Liquid-Cooled Workstations, Enterprise AI Servers
99 tracked101 tracked · +Liquid-Cooled Workstations, Enterprise AI Servers
source - capability
- version
- capability
New capabilities: C++ SDK, Python API, MLIR Compiler
95 tracked98 tracked · +C++ SDK, Python API, MLIR Compiler
source - capability
New capabilities: Enterprise-class AI servers
94 tracked95 tracked · +Enterprise-class AI servers
source - version
- model
Tenstorrent Blackhole added Gemma4, GLM-5.2, Kimi-K3 to its model stack
Gemma-2, Llama-3.3-70B, Qwen3-32B, Tensix cores, tt-metal / tt-nnGLM-5.2, Gemma-2, Gemma4, Kimi-K3, Llama-3.3-70B, Qwen3-32B, Tensix cores, tt-metal / tt-nn
source - capability
- version
- capability
New capabilities: PCIe AI Processor, C++ SDK / Metalium, PyTorch/JAX/ONNX Support
89 tracked92 tracked · +PCIe AI Processor, C++ SDK / Metalium, PyTorch/JAX/ONNX Support
source - version
- capability
- capability
- capability
New capabilities: PCIe card form factor, Enterprise-class AI server options
85 tracked87 tracked · +PCIe card form factor, Enterprise-class AI server options
source - version
- capability
New capabilities: Liquid-cooled workstation support, PCIe Card support
83 tracked85 tracked · +Liquid-cooled workstation support, PCIe Card support
source - capability
- version
- capability
New capabilities: AI Hardware, Liquid Cooled Workstation
80 tracked82 tracked · +AI Hardware, Liquid Cooled Workstation
source - version
- capability
New capabilities: multi-server configurations, liquid cooled workstation, low-level kernels
75 tracked80 tracked · +multi-server configurations, liquid cooled workstation, low-level kernels, low-level C++ SDK, high-level Python API
source - version
- model
Tenstorrent Blackhole added Gemma-2, Llama-3.3-70B, Qwen3-32B to its model stack
Tensix cores, tt-metal / tt-nnGemma-2, Llama-3.3-70B, Qwen3-32B, Tensix cores, tt-metal / tt-nn
source - capability
New capabilities: Open-source software stack, PCIe cards, TT-Metalium SDK
72 tracked75 tracked · +Open-source software stack, PCIe cards, TT-Metalium SDK
source - version
- capability
New capabilities: Liquid-cooled Workstations
71 tracked72 tracked · +Liquid-cooled Workstations
source - version
- capability
New capabilities: AI hardware, Workstation, Server
68 tracked71 tracked · +AI hardware, Workstation, Server
source - capability
New capabilities: liquid-cooled workstation, LLM inference
66 tracked68 tracked · +liquid-cooled workstation, LLM inference
source - capability
New capabilities: AI accelerator, PCIe card, Workstation AI
62 tracked66 tracked · +AI accelerator, PCIe card, Workstation AI, Enterprise-class AI server
source - capability
New capabilities: Tensor Core / Tensix cores
61 tracked62 tracked · +Tensor Core / Tensix cores
source - capability
New capabilities: AI hardware accelerator, Liquid-cooled options, PCIe Card compatibility
58 tracked61 tracked · +AI hardware accelerator, Liquid-cooled options, PCIe Card compatibility
source - version
- capability
Tenstorrent Blackhole compared
Straight head-to-head pages against the busiest products in Chips & Compute.
How to cite this page
Free to cite and reuse under CC BY 4.0. Permalink: https://tomorrow.aliensquad.ai/tools/tenstorrent
APA
Tomorrow. (2026). Tenstorrent Blackhole — version, pricing and model stack [Data set entry]. AlienSquad. Retrieved 2026-09-17, from https://tomorrow.aliensquad.ai/tools/tenstorrent
BibTeX
@misc{tomorrow-tools-tenstorrent,
author = {{Tomorrow}},
title = {Tenstorrent Blackhole — version, pricing and model stack},
year = {2026},
publisher = {AlienSquad},
howpublished = {\url{https://tomorrow.aliensquad.ai/tools/tenstorrent}},
note = {Accessed: 2026-09-17}
}