Kimi K3.1 Pro Soft-Derisked is now on Vani
Kimi K3.1 Pro with Vani's soft-derisked instructions: fewer refusals than standard Kimi K3.1 Pro, more conservative than GLM 5.3 Derisked. For adults, in Vani Chat on Plus and higher and on the Vani API.
Illustration
Kimi K3.1 Pro Soft-Derisked is available today in Vani Chat on Plus and higher, and on the Vani API as moonshotai/kimi-k3.1-pro-soft-derisked.
It's the same Kimi K3.1 Pro model from Moonshot AI, running with Vani's own soft-derisked instructions. It refuses less often than standard Kimi K3.1 Pro, and it's more conservative than GLM 5.3 Derisked.
Where it fits
| Model | Refusals |
|---|---|
| Kimi K3.1 Pro | Standard |
| Kimi K3.1 Pro Soft-Derisked | Fewer |
| GLM 5.3 Derisked | Fewest |
What stays the same
- It always thinks first. Pick low, high or max effort. Vani shows a live "Thinking" timer while it works.
- 1 million tokens of context and up to 131,072 tokens of output.
- In Vani Chat, it uses your allowance at the same rate as Kimi K3.1 Pro.
For adults
Like GLM 5.3 Derisked, it's for adults. The first time you pick it in Vani Chat, Vani asks you to confirm you're 18 or older. Vani's terms still apply.
Use it on the API
The model ID is moonshotai/kimi-k3.1-pro-soft-derisked. It works with the OpenAI-compatible Chat Completions and Responses APIs and the Anthropic-compatible Messages API, with streaming and tool calling.
| Per 1M tokens | Price |
|---|---|
| Input | $5.00 |
| Cached input | $0.50 |
| Output | $7.00 |
Vani's instructions add about 1,100 input tokens to each request. They're usually billed at the cached rate.
Prepaid API keys from platform.vani.ai and keys created in Vani Chat on Plus or higher can use it today.