AMD Token Lab gives you a free API key for DeepSeek V4.1 Flash and other open models for 30 days
Free Token Lab API key with 10 credits for DeepSeek, Qwen, Gemma and other open models over 30 days
AMD has launched Token Lab in its global AI Developer Program. Join for free and you get an API key with 10 credits to spend over 30 days on open models such as DeepSeek-V4.1-Flash, Qwen3.8-27B and Gemma-4-31B-IT, through an OpenAI-compatible endpoint.
- Eligible plans
- AMD Token Lab (Beta)
- Eligibility
- AMD AI Developer Program (Global) members; joining is free and no card is needed
- Benefit duration
- 30 days from key creation, with a one-time 15-day extension available in the last 5 days
The briefing
AMD has added Token Lab to its global AI Developer Program. Members can get a free API key and call a set of open models served on AMD GPUs. The endpoint is OpenAI-compatible, works with the Anthropic SDK too, and no card is needed.
What you get
Each API key comes with 10 credits and is valid for 30 days from the day you create it. Credits do not reset daily. Each model charges its own rate per million tokens: DeepSeek-V4.1-Flash costs 0.28 in and 0.56 out, DeepSeek-V4-Flash 0.14 in and 0.28 out, and Qwen3.8-27B, Gemma-4-31B-IT and the MiniCPM models just 0.035 in and 0.07 out. Counting input only, 10 credits cover roughly 35 million tokens on DeepSeek-V4.1-Flash, or close to 300 million on Qwen3.8-27B.
The catalog also includes d3 and Clef, two decision models that answer typed questions with probabilities. Each key is limited to 20 requests per minute, so it suits development and testing better than heavy batch jobs.
Requests are rejected once your credits run out or the 30 days end. When you have 2 credits or fewer left, or you are in the last 7 days of your access period, you can request more credits from the page. In the last 5 days you can also extend access once by 15 days.
How to use
- Open AMD Token Lab and sign in with an AMD AI Developer Program (Global) account. If you are not a member yet, join first; it is free.
- Click Get my API key to create a key starting with tl-.
- Set your client's base URL to https://api.tokenlab.amd.com/v1, paste the key, and use the model name shown on each card, for example DeepSeek-V4.1-Flash-1017.
See the AMD AI Developer Program for the program's other member perks.