MiniMax M3.1 Flash Is Free on BlockRun, No Wallet

nvidia/minimax-m3.1-flash is live on BlockRun as of today, November 5, on Base and on Solana. It costs nothing, and you don't need a key, a wallet or an account to call it.

Model	Input / 1M	Output / 1M	Context
`nvidia/minimax-m3.1-flash`	Free	Free	1M

It reasons before it answers and it calls tools. Send it through the same OpenAI-compatible endpoint as every other model.

What it is for

Agent loops on a budget. Prototype a tool-calling agent and run it as often as you like before you pay for a larger model.
Long inputs. With a 1M-token window, a large document or a codebase fits in the prompt.
Work that benefits from reasoning. It returns its reasoning in reasoning_content, so you can log it or show it without parsing the answer.

What we checked before listing it

We don't list a model on a 200 from a models endpoint. Before it went into the catalog, we sent it real requests:

a 1,725-token prompt, three times, answered in 2.4 to 6.3 seconds
a streamed completion, with usage at the end
a tool call, where the model chose to call the tool

We also sent it an image and got back an empty answer, so we did not list it as a vision model.

What free means here

NVIDIA hosts this model on its free tier, and those terms allow prompts to be used to improve the service. Don't send it anything sensitive. For private work, choose a paid model.

Free capacity is shared, so a busy period can return a 429, or another free model may answer in its place. The response headers name the model that answered. If you need a specific model every time, choose a paid one.

Calling it

curl https://blockrun.ai/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nvidia/minimax-m3.1-flash",
    "messages": [{"role": "user", "content": "Explain what this regex matches: ^[a-z]+@[a-z]+\\.com$"}]
  }'

On Solana, make the same call through sol.blockrun.ai.