Skip to content

New: DeepSeek V4.1 Flash Derisked. In chat and on the API.

See modelsWhat's newReport a bug
Open chat

Derisked models: fewer refusals for legitimate work.

Most models are trained for a consumer audience, so they decline work that is perfectly legitimate. A de-risked model is a modified version of the original: its weights have been edited to take out that trained reflex to say no, so it refuses far less.

For adults (18+). Use is governed by our Terms of Service, which prohibit illegal and harmful use.

DeepSeek V4.1 Flash Derisked$1.30 / $2.60GLM 5.3 Derisked$3.90 / $6.50Kimi K3.1 Pro Soft-Derisked$5.00 / $7.00
De-risked, 18+
Size and colour: output price
Under each red name: input / output per 1M tokens

Available now

Every de-risked model in the live catalogue, with its API price. They are marked in red in chat, on pricing and in the API model list.

DeepSeek

DeepSeek V4.1 Flash Derisked

De-risked
Context window
1M context
Thinking
Adjustable thinking
Plans
All plans

Fewer refusals (18+): follows your instructions more closely. Vani's usage rules still apply.

Best for
Fiction for adults · Frank answers
Good to know
  • For adults (18+) only
  • Always thinks before answering, so replies use more output
  • Much heavier on your allowance than standard models
API model id
deepseek/deepseek-v4.1-flash-derisked
Input, per 1M tokens
$1.30
Output, per 1M tokens
$2.60

Prepaid API prices. In Vani Chat it uses your plan's allowance: tokens per plan.

Z.AI

GLM 5.3 Derisked

De-risked
Context window
1M context
Thinking
Adjustable thinking
Plans
All plans

Fewer refusals (18+): follows your instructions more closely. Vani's usage rules still apply.

Best for
Fiction for adults · Frank answers
Good to know
  • For adults (18+) only
  • Much heavier on your allowance than standard models
API model id
zhipu/glm-5.3-derisked
Input, per 1M tokens
$3.90
Output, per 1M tokens
$6.50

Prepaid API prices. In Vani Chat it uses your plan's allowance: tokens per plan.

Moonshot AI

Kimi K3.1 Pro Soft-Derisked

De-risked
Context window
1M context
Thinking
Adjustable thinking
Plans
Plus and above

Fewer refusals than standard Kimi K3.1 Pro, more conservative than GLM 5.3 Derisked. It's Kimi K3.1 Pro with Vani's soft-derisked instructions. For adults (18+).

Best for
Long-context research and analysis · Coding
Good to know
  • Still refuses some requests; for the fewest refusals use GLM 5.3 Derisked
  • The system prompt adds about 1,100 input tokens per request
API model id
moonshotai/kimi-k3.1-pro-soft-derisked
Input, per 1M tokens
$5.00
Output, per 1M tokens
$7.00
Cached input
$0.50

Prepaid API prices. In Vani Chat it uses your plan's allowance: tokens per plan.

What de-risked actually means.

  1. A modified model, not a filter switched off

    Refusals are trained into a model's weights; there is no separate filter to remove. De-risking edits the weights to take that behaviour out. The edit is aimed at refusals, not at knowledge or skill, so the model is built to stay as capable as the original. It is still a changed model, so its answers can differ from the original's.

  2. Fewer refusals, same rules

    No consumer refusal layer sits between your prompt and the model, so legitimate work is answered instead of declined on reflex. A de-risked model refuses far less, not never: it can still decline a request now and then. Our Terms of Service still apply.

  3. The judgement is yours

    You decide what's appropriate for your users and your jurisdiction, rather than inheriting a provider's view of it. That responsibility comes with the access.

Work that consumer-tuned models often refuse

  • Security research

    Often refused: Refuses to analyse malware or reason about exploits.

    With a de-risked model: Reverse-engineer samples, write detection rules, build red-team tooling, and walk an exploit chain end to end.

  • Medical and legal

    Often refused: Buries every answer under liability disclaimers.

    With a de-risked model: Clinical specifics, drug interactions, dosage reasoning, and case detail — answered directly, for professionals who need the substance.

  • Fiction and narrative

    Often refused: Declines violence, mature themes, and morally grey characters.

    With a de-risked model: Write crime, horror, war, and adult drama with the darkness the story needs, instead of a sanitised rewrite.

  • Finance and research

    Often refused: Hedges on anything that resembles advice or a position.

    With a de-risked model: Direct risk analysis, contrarian positions, and research into sensitive or politically charged topics without the disclaimer tax.

What still applies

Vani applies no content-policy filtering to de-risked models. Responsibility for how you deploy them sits with you.

  • Our Terms of Service apply to de-risked models exactly as to every other model. They prohibit anything illegal under applicable law, including the generation or distribution of child sexual abuse material, and using outputs to cause physical harm, to target individuals for harassment, or to facilitate violence.
  • De-risked models are for adults (18+), and they are always labelled, so choosing one is deliberate.

Questions

What does de-risked mean?
Most models are trained for a consumer audience, so they decline work that is perfectly legitimate. A de-risked model is a modified version of the original: its weights have been edited to take out that trained reflex to say no, so it refuses far less. Our Terms of Service still apply.
Is a de-risked model less capable than the standard one?
Refusals are trained into a model's weights; there is no separate filter to remove. De-risking edits the weights to take that behaviour out. The edit is aimed at refusals, not at knowledge or skill, so the model is built to stay as capable as the original. It is still a changed model, so its answers can differ from the original's.
What is still not allowed?
Everything our Terms of Service prohibit, on every model: anything illegal under applicable law, including child sexual abuse material, and using outputs to cause physical harm, to target individuals for harassment or to facilitate violence. De-risked models are for adults (18+).
Is a de-risked model the same as an uncensored model?
People often look for it under that name. A de-risked model declines far fewer legitimate requests, because its trained refusal behaviour has been removed, but it is not a way around the law or our rules: the Terms of Service still apply, and you are responsible for how you use what it generates.
What does Soft-Derisked mean?
Kimi K3.1 Pro Soft-Derisked: fewer refusals than standard Kimi K3.1 Pro, more conservative than GLM 5.3 Derisked. It's Kimi K3.1 Pro with Vani's soft-derisked instructions. For adults (18+).
How much do de-risked models cost?
On the prepaid Vani API, DeepSeek V4.1 Flash Derisked costs $1.30 per 1M input tokens and $2.60 per 1M output tokens; GLM 5.3 Derisked costs $3.90 per 1M input tokens and $6.50 per 1M output tokens; Kimi K3.1 Pro Soft-Derisked costs $5.00 per 1M input tokens and $7.00 per 1M output tokens ($0.50 per 1M cached input tokens). In Vani Chat they use your plan's allowance; the pricing page shows how far each plan goes with each model.
Can I try one without an account?
Yes. GLM 5.3 Derisked is in guest chat: open it and pick the model, no sign-up needed. A free account gives you your own allowance and saved chats.
How do I use a de-risked model from the API?
Send deepseek/deepseek-v4.1-flash-derisked or zhipu/glm-5.3-derisked or moonshotai/kimi-k3.1-pro-soft-derisked as the model to the OpenAI- or Anthropic-compatible endpoints, with a key from the Vani API Platform.

Fewer refusals, same rules

For adults (18+). Use is governed by our Terms of Service, which prohibit illegal and harmful use.