AgentKernelArena compatibility matrix

AgentKernelArena compatibility matrix#

2026-07-20

2 min read time

Applies to Linux

Supported and tested hardware, Docker images, software versions, agent CLIs, and model providers for AgentKernelArena.

The following hardware configurations are supported and tested.

AMD Instinct GPU	ROCm version	Notes
MI300X	7.2 (Bundled in the selected SGLang image.)	`target_gpu_model: MI300X`
MI325X	7.2 (Bundled in the selected SGLang image.)	`target_gpu_model: MI325X`
MI355X	7.2 (Bundled in the selected SGLang image.)	`target_gpu_model: MI355X`

The following software versions are required or verified.

Component	Version	Notes
Linux	Ubuntu 22.04, Ubuntu 24.04
hipcc	Matches ROCm image	Required for HIP tasks.
rocprof-compute	Matches ROCm image	Required for HIP performance profiling.
Docker	Current stable release	Required; serial experiments run through `make docker-run`; multi-GPU experiments run through `make docker-parallel-run`.
SGLang runtime image	`lmsysorg/sglang:v0.5.12-rocm720-mi30x` for `gfx942`; `lmsysorg/sglang-rocm:v0.5.14-rocm720-mi35x-20260705` for `gfx950`	The verified `gfx950` digest is `sha256:b435b508b5aa696abb25c909341ce73e41574c4271cf716bed72418dcea86b78`. Override with `AKA_DOCKER_IMAGE`, `AKA_DOCKER_IMAGE_GFX942`, or `AKA_DOCKER_IMAGE_GFX950`.
Python	Provided by the image (for example, 3.10)	Bundled in the SGLang image.
Node.js and npm	Node.js 22 with a current npm	Required on the host only for the alternative npm installation of Claude Code or another npm-installed agent CLI.
PyTorch	ROCm build bundled in the image	Provided by the SGLang Docker image.
Triton	Bundled with the image’s ROCm PyTorch	Required for Triton task categories.
AITER	`0.1.17.dev110+g9127c94a1` in the verified `gfx950` image	Required by AITER-backed task oracles and kernels.
FlyDSL	`0.2.2` in the verified `gfx950` image (or `make docker-setup-flydsl` when absent)	Required for `flydsl2flydsl`, `torch2flydsl`, and `triton2flydsl` tasks.

The following templates are selectable in the current AgentType registry. See Install AgentKernelArena and Configure agents and models for setup instructions.

Template	Runtime dependency
`cursor`	Cursor Agent CLI and host login state.
`claude_code`	Native/local or npm-installed Claude Code CLI and host login state.
`codex`	Codex CLI and host login state.
`geak_v3`	GEAK CLI; HIP-oriented integration.
`geak_v3_triton`	GEAK CLI; Triton-oriented integration.
`mini_swe_triton`	mini-swe-agent/GEAK dependencies.
`task_validator`	Claude Code or Codex backend configured in `agents/task_validator/agent_config.yaml`.

Model/provider support is integration-specific; run configuration files do not configure a provider.

Provider	Notes
OpenAI	Use a selected integration or CLI configured for OpenAI.
Anthropic	Use a selected integration or CLI configured for Anthropic.
OpenRouter or another OpenAI-compatible service	Supported when the selected integration accepts a custom provider/base URL.
Local vLLM	`make vllm` starts an OpenAI-compatible endpoint on port `30001`; configure the selected integration to use it.