Uncensored models for security research.
DeepSeek V4.1 Flash and GLM 5.3 without refusals, for pentesting, red teaming, malware analysis and code. Change one base URL and your coding agent just works.
Built for people who break things for a living.
An API that answers the questions security work actually requires, with the controls a team needs to run it safely.
Drop-in compatible
Speaks the Anthropic Messages and OpenAI Chat Completions formats, with streaming and tool calls.
Zero data retention
Prompts and responses live only in memory. We store token counts, never content.
Per-key controls
Spend limits, expirations, model restrictions and one-click rotation for every key.
Live usage
See spend, tokens and requests per key as they happen. No surprises on your balance.
Two open-weight models, no refusals.
Prices in USD per million tokens. Cached input is billed at the cache-read rate. Reasoning tokens count as output.
DeepSeek V4.1 Flash 1017 1M
deepseek-v4-1-flash-1017-1mLong-context MoE model for coding agents and large codebases. The default for Claude Code.
GLM 5.3 Uncensored
glm-5-3-uncensoredStrong reasoning model for analysis, report writing and exploit reasoning.
Pick a plan. Its credits arrive every month.
Every plan includes API access to all models and features, plus credits to spend on tokens each month. Bigger plans include more credits per dollar.
- Access to every model and feature
- Up to 10 API keys with spend limits and expirations
- Unused credits roll over to the next month
- Add extra credits from $10 whenever you need them
- Requests stop at $0, so you can't overspend
- Upgrade, downgrade or cancel any time from the console
What a dollar buys
DeepSeek V4.1 Flash| Workload | Cost | Per $10 |
|---|---|---|
| Chat message 2K in, 500 out | $0.024 | ~400 |
| Agent step, cached 60K context, 1K out | $0.10 | ~100 |
| Agent step, uncached 60K in, 1K out | $0.50 | ~20 |
Uncensored doesn't mean anything goes.
Every account agrees to our acceptable use policy.
Built for
- Authorized penetration testing and red teaming
- CTFs, labs and security training
- Malware analysis and incident response
- Vulnerability research with responsible disclosure
Gets you banned
- Attacking systems without authorization
- Malware, fraud or phishing against real people
- Any sexual content involving minors
- Biological, chemical or nuclear weapons
Questions developers ask.
What does "uncensored" mean here?
The models' refusal behavior has been removed at the weight level. They answer technical security questions that commercial models usually block.
Does it work with Claude Code?
Yes. Set ANTHROPIC_BASE_URL to the RedLine gateway and ANTHROPIC_API_KEY to your key. Claude Code's default model names are routed to DeepSeek V4.1 Flash automatically.
Do you store my prompts?
No. Prompts and responses exist only in memory while a request runs. We keep token counts and cost for billing. See the Zero Data Retention Policy.
How does billing work?
Pick a monthly plan: Starter ($50, includes $30 in credits), Pro ($100, includes $75) or Scale ($250, includes $200). Each monthly charge adds the plan's credits to your balance, and tokens are paid from it. If you run out before the month ends, add more credits at any amount. Unused credits roll over.
Can I change plans?
Yes, from the console. An upgrade is charged right away, adds the new plan's credits at once and starts a new monthly period. A downgrade takes effect when your current month ends.
What happens when I run out of credits?
Requests return HTTP 402 until you add credits. You are never charged beyond your balance.
Can I cancel?
Yes, any time from the console. You keep access until the end of the month you paid for, and unused credits stay on your account if you come back.
Can I limit what a key can spend?
Yes. Every key can have its own spend limit, expiration date and allowed models, and you can rotate or revoke it at any time.
Start in two minutes.
Create an account, add credits and point your client at RedLine.