Intel GPU AI Skills is a collection of agent skills for setting up, running, benchmarking, and profiling Hugging Face models on Intel GPUs. It supports workflows involving PyTorch, vLLM-XPU, SGLang-XPU, llama.cpp-SYCL, and migration from CUDA to XPU. The catalogue contains the project's skills, instructions, agent, and plugin.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add intel/gpu-ai-skills --skill torch-xpu-benchgit clone --depth 1 https://github.com/intel/gpu-ai-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/intel/gpu-ai-skills/torch-xpu-bench)<a href="https://agentmods.dev/skills/intel/gpu-ai-skills/torch-xpu-bench"><img src="https://agentmods.dev/badge/skills/intel/gpu-ai-skills/torch-xpu-bench/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/intel/gpu-ai-skills/torch-xpu-bench"><img src="https://agentmods.dev/badge/skills/intel/gpu-ai-skills/torch-xpu-bench.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00096 | $0.02054 |
| Opus 5 | $0.00048 | $0.01027 |
| Sonnet 5 | $0.00019 | $0.00411 |
| Haiku 4.5 | $0.00010 | $0.00205 |
Grade A, and why
torch-xpu-bench scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 198 lines — stays where its author put it; the contents beside it link to each section on GitHub.
torch-xpu-bench
Pure-PyTorch benchmarks of HF models on XPU. Measures bare
model.generate() with no scheduler, no continuous batching, no
API overhead. For serving-shape numbers (TTFT/TPOT/ITL under
concurrency) use vllm-xpu-bench.
Important: This skill provides a ready-to-use benchmark script at
{base_dir}/scripts/bench.py. Do not write your own benchmark script;
use the provided one.
Use cases:
- TTFT and decode rate of a single sequence at a given dtype.
- A/B dtypes (bf16 vs fp16), compile on/off, model size.
- Confirm
--enforce-eageris hurting in vLLM before flipping it. - Validate model-can-it-fit predictions against
max_memory_allocated.
For diffusion and encoder-only models, see
references/non-llm-snippets.md.
CUDA → XPU translation
| CUDA | XPU |
|---|---|
torch.cuda.synchronize() |
torch.xpu.synchronize() |
torch.cuda.Event(enable_timing=True) |
torch.xpu.Event(enable_timing=True) |
torch.cuda.max_memory_allocated() |
torch.xpu.max_memory_allocated() |
torch.cuda.reset_peak_memory_stats() |
torch.xpu.reset_peak_memory_stats() |
torch.cuda.empty_cache() |
torch.xpu.empty_cache() |
Two gotchas:
- Synchronise before timing. XPU kernels launch async; without
torch.xpu.synchronize()you measure launch latency. - Warm up. First call compiles XPU Triton kernels; first compiled-graph call triggers capture. Discard the first 1–3 runs.
Hosts with non-Intel GPUs
If nvidia-smi or rocm-smi reports hardware, bench code that
doesn't pin a device may silently land on CUDA/ROCm. Defences:
export CUDA_VISIBLE_DEVICES= # hide all NVIDIA
export HIP_VISIBLE_DEVICES= # hide all AMD ROCm
export ZE_AFFINITY_MASK=0 # pick a specific Intel device
scripts/bench.py asserts model parameters land on xpu after
from_pretrained; replicate in your own benches. Cross-check with
xpu-smi dump -d 0 -m 5 — if XPU memory doesn't climb, the bench
is on the wrong device.
These all load onto xpu:0 when only an Intel XPU is visible:
device_map="xpu" / "xpu:0" / 0 / "auto" / {"": "xpu"} / {"": 0}.
hf_device_map is empty for whole-model placement; populates only
on multi-GPU splits.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 198 lines · 96 tokens per session scan A fcc69d2680fd
torch-xpu-bench is a skill published in the GitHub repository intel/gpu-ai-skills (21 stars, last pushed 4d ago), licensed Apache-2.0. It adds 96 tokens to every session and 2,054 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
agent-platform-rag-engine-management
Manage and query Agent Platform RAG Engine Corpora and retrieve grounded contexts using the Google GenAI SDK. Use when listing RAG corpora or files, inspecting a corpus, retrieving contexts, or generating content grounded in a RAG corpus. Do not use for standard database queries (use SQL/Spanner skills), Google…
agent-platform-model-registry
Agent Platform Model Registry Management. Use when you need to upload, list, describe, update, or delete machine learning models (and their versions) in the Agent Platform Model Registry. Don't use for model training, model deployment to endpoints, or managing non-Agent Platform models.
foundry-config-setup
Resolve missing setup caused by a hardcoded Foundry project endpoint or model in a sample. Use when a sample fails because it uses a placeholder/hardcoded projectendpoint (for example "https://your-project.services.ai.azure.com") or a hardcoded model instead of reading them from the environment.
google-cloud-solution-agentic-analytics-spark-knowledge-catalog
Discovers requirements and generates guidance to design and deploy a governed, secure agentic-analytics solution for data that's distributed across Google Cloud, other cloud providers, or on-premises. Data that's outside Google Cloud (such as data from Databricks, Snowflake, Salesforce, SAP, or Oracle systems) is…
training-check
Interactively monitor training metrics from the current Codex session, periodically checking WandB or fallback logs for NaN, divergence, plateaus, and broken runs.
nemo-automodel-launcher-config
Configure NeMo AutoModel job launches for interactive runs, Slurm clusters, and SkyPilot cloud execution.