Playground
Ask it the thing you would not ask a vendor.
Send our endpoint what you actually need.
qwen-flash-next
uncensored · FP8
session · 9f3c-2a71
qwen-flash-next · xhigh · uncensored
Session open. qwen-flash-next · cybersecurity build · reasoning xhigh.
Ask anything, or take one of the chips under the composer. Effort is per request: off when you want the first token fast, low as a working default, xhigh when the answer has to hold together.
try one
- context262,144 tokens
- max output131,072 tokens
- precisionFP8
- concurrency5
- retentionZDR · 0 bytes
Point your own client at it
One command, then it is your client's problem.
The endpoint implements the OpenAI chat-completions shape, so anything that accepts a custom base URL works unchanged: an SDK, a script, a roleplay frontend, a notebook. There is no SDK to install and no model picker — the model id is constant.
export KEY="sk-rwi-…"
curl https://api.rawinfer.com/v1/chat/completions \
-H "Authorization: Bearer $KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen-flash-next",
"reasoning_effort": "off",
"temperature": 0.7,
"stream": true,
"messages": [
{"role": "user", "content": "A path parameter is joined to a directory and served; the check only rejects the literal '..'. How would an attacker exploit this, and how do you fix it?"}
]
}'Three things to know before you start
- The model id never changes. qwen-flash-next is the only value /v1/models returns. If a tutorial asks you to select a model, select that one.
- Effort is a request parameter. Set reasoning_effort to off for throughput, xhigh when the answer has to hold together, low in between.
- Nothing is stored. Prompts and completions are never written to disk, never logged, never trained on. If you need a history, keep it yourself.
Crypto only. Try it for $5 — a hundred requests, no expiry.
One model · one endpoint · one key
Point your own client at the model.
Fifty a month for unlimited at concurrency 5. Try it for $5 if you want a shorter look. Crypto only.