Free Beats Paid: Gemma 4 31B on AMD Token Lab Launches Local Chat Page and ADK Agent
agents gemma openai
| Source: Dev.to | Original article
AMD Token Lab offers a free OpenAI‑compatible API on AMD GPUs, featuring a local chat page, terminal client, and an ADK agent that runs Gemma 4 31B.
AMD has launched Token Lab, a free, OpenAI‑compatible API that runs on AMD GPUs and is bundled with the AMD AI Developer Program. The service lets developers call the Gemma 4 31B model through a lightweight ADK agent, while a local chat page stores the API key on the user’s machine rather than in the browser. A terminal client is also provided for command‑line access.
The move lowers the cost barrier for developers who want to experiment with large‑scale language models on AMD hardware. By offering a no‑charge endpoint that mirrors OpenAI’s interface, AMD aims to attract projects that might otherwise rely on commercial cloud providers. Keeping the key off the browser addresses a common security concern, and the inclusion of a ready‑made agent simplifies integration into emerging AI‑driven workflows such as autonomous assistants and tool‑using bots.
The announcement arrives amid a surge of interest in AI agents, following recent coverage of Claude’s Duet, ChatGPT’s new Wallet feature and broader discussions about long‑horizon decision models. AMD’s free API could accelerate the adoption of locally hosted agents and give developers a viable alternative to cloud‑only solutions.
Watch for early adopters’ performance benchmarks, the extent to which the ADK agent is incorporated into third‑party tools, and any updates to the AMD AI Developer Program that might expand the offering. The rollout also raises questions about how AMD will compete with larger cloud players that are intensifying data collection and service integration efforts.
Sources
Back to AIPULSEN