[Dagon Inference](https://inference.dagon.so/) / Model changelog

# Model changelog

Published 2026-11-08 · Last updated 2026-11-09

What changed in the [model catalog](https://inference.dagon.so/abliterated-llm-pricing), newest first: models added or removed, renamed ids, refusal labels, providers, and prices that moved by 5% or more. Follow it with the [Atom feed](https://inference.dagon.so/changelog.xml) or read [/changelog.json](https://inference.dagon.so/changelog.json); the live list is [`GET /v1/models`](https://inference.dagon.so/v1/models).

## 2026-11-09

- **New provider for GLM-5.3 Flash: Cheaper Inference** GLM-5.3 Flash (glm-5.3-flash): now also served by Cheaper Inference. [Model page](https://inference.dagon.so/models/glm-5.3-flash)
- **New model: Kimi K3.1 Pro** New model: Kimi K3.1 Pro (kimi-k3.1-pro), served by Refuseless and Silk Compute, with standard refusal behavior. $0.275 input, $0.165 cached input, $2.75 output per 1M tokens. [Model page](https://inference.dagon.so/models/kimi-k3.1-pro)
- **New model: GLM-5.3 Flash** New model: GLM-5.3 Flash (glm-5.3-flash), served by Lutheos, with standard refusal behavior. $0.088 input, $0.0374 cached input, $0.462 output per 1M tokens. [Model page](https://inference.dagon.so/models/glm-5.3-flash)
- **New model: DeepSeek V4.1 Flash** New model: DeepSeek V4.1 Flash (deepseek-v4.1-flash), served by Silk Compute, with standard refusal behavior. $0.022 input, $0.0044 cached input, $0.176 output per 1M tokens. [Model page](https://inference.dagon.so/models/deepseek-v4.1-flash)
- **New model: MiniMax M3.1 Flash** New model: MiniMax M3.1 Flash (minimax-m3.1-flash), served by Refuseless, with standard refusal behavior. $0.022 input, $0.0022 cached input, $0.088 output per 1M tokens. [Model page](https://inference.dagon.so/models/minimax-m3.1-flash)
- **New model: DeepSeek V4.1 Flash Fast** New model: DeepSeek V4.1 Flash Fast (deepseek-v4.1-flash-fast), served by Refuseless, with standard refusal behavior. $0.011 input, $0.0022 cached input, $0.088 output per 1M tokens. [Model page](https://inference.dagon.so/models/deepseek-v4.1-flash-fast)

## 2026-11-08

- **Renamed: kimi-k3.1-pro is now kimi-k3.1-pro-abliterated** Kimi K3.1 Pro Abliterated has a new model id: kimi-k3.1-pro-abliterated (was kimi-k3.1-pro). Requests must use the new id; the old one returns 404 model\_not\_found. [Model page](https://inference.dagon.so/models/kimi-k3.1-pro-abliterated)
- **Refusal label removed: Kimi K3.1 Pro Abliterated** Kimi K3.1 Pro Abliterated (kimi-k3.1-pro): refusal label changed from standard to removed (refusal behavior removed). [Model page](https://inference.dagon.so/models/kimi-k3.1-pro)
- **Renamed: Kimi K3.1 Pro is now Kimi K3.1 Pro Abliterated** kimi-k3.1-pro is now named Kimi K3.1 Pro Abliterated (was Kimi K3.1 Pro). The model id is unchanged. [Model page](https://inference.dagon.so/models/kimi-k3.1-pro)

## 2026-11-07

- **Price rise: Abliterated Large v2** Abliterated Large v2 (abliterated-model-large-v2): input $1.10 to $3.30, cached input $0.165 to $0.33, output $2.20 to $5.50 per 1M tokens, markup included. [Model page](https://inference.dagon.so/models/abliterated-model-large-v2)
- **Providers changed: Abliterated Large v2** Abliterated Large v2 (abliterated-model-large-v2): no longer served by Experiential Labs, Refuseless and Silk Compute. [Model page](https://inference.dagon.so/models/abliterated-model-large-v2)
- **New model: GLM-5.3 Abliterated** New model: GLM-5.3 Abliterated (glm-5.3-abliterated), served by Experiential Labs, Refuseless and Silk Compute, with refusal behavior removed. $1.10 input, $0.165 cached input, $2.20 output per 1M tokens. Its providers served it as abliterated-model-large-v2 before. [Model page](https://inference.dagon.so/models/glm-5.3-abliterated)

## 2026-11-06

- **New model: Kimi K3.1 Pro** New model: Kimi K3.1 Pro (kimi-k3.1-pro), served by Silk Compute, with standard refusal behavior. $4.40 input, $0.55 cached input, $22.00 output per 1M tokens. [Model page](https://inference.dagon.so/models/kimi-k3.1-pro)
- **New model: Hy4** New model: Hy4 (hy4), served by Silk Compute, with standard refusal behavior. $1.76 input, $0.33 cached input, $3.52 output per 1M tokens. [Model page](https://inference.dagon.so/models/hy4)
- **New provider for GLM-5.3 Flash Abliterated: Silk Compute** GLM-5.3 Flash Abliterated (glm-5.3-flash-abliterated): now also served by Silk Compute. [Model page](https://inference.dagon.so/models/glm-5.3-flash-abliterated)
- **New provider for DeepSeek V4.1 Flash Abliterated: Silk Compute** DeepSeek V4.1 Flash Abliterated (deepseek-v4.1-flash-abliterated): now also served by Silk Compute. [Model page](https://inference.dagon.so/models/deepseek-v4.1-flash-abliterated)
- **New provider for Abliterated Large v2: Silk Compute** Abliterated Large v2 (abliterated-model-large-v2): now also served by Silk Compute. [Model page](https://inference.dagon.so/models/abliterated-model-large-v2)

## 2026-11-05

- **Price cut: DeepSeek V4.1 Flash Abliterated** DeepSeek V4.1 Flash Abliterated (deepseek-v4.1-flash-abliterated): input $0.132 to $0.088, cached input $0.0275 to $0.0176, output $0.495 to $0.352 per 1M tokens, markup included. [Model page](https://inference.dagon.so/models/deepseek-v4.1-flash-abliterated)
- **New provider for DeepSeek V4.1 Flash Abliterated: Lutheos** DeepSeek V4.1 Flash Abliterated (deepseek-v4.1-flash-abliterated): now also served by Lutheos. [Model page](https://inference.dagon.so/models/deepseek-v4.1-flash-abliterated)

## 2026-11-02

- **Price cut: GLM-5.3 Flash Abliterated** GLM-5.3 Flash Abliterated (glm-5.3-flash-abliterated): input $0.165 to $0.132, cached input $0.066 to $0.044 per 1M tokens, markup included. [Model page](https://inference.dagon.so/models/glm-5.3-flash-abliterated)
- **Price cut: Abliterated Large v2** Abliterated Large v2 (abliterated-model-large-v2): input $1.375 to $1.10, cached input $0.275 to $0.165, output $2.475 to $2.20 per 1M tokens, markup included. [Model page](https://inference.dagon.so/models/abliterated-model-large-v2)
- **Price cut: DeepSeek V4.1 Flash Abliterated** DeepSeek V4.1 Flash Abliterated (deepseek-v4.1-flash-abliterated): input $0.242 to $0.132, cached input $0.0495 to $0.0275, output $0.825 to $0.495 per 1M tokens, markup included. [Model page](https://inference.dagon.so/models/deepseek-v4.1-flash-abliterated)

## 2026-11-01

- **New model: DeepSeek V4.1 Flash Abliterated** New model: DeepSeek V4.1 Flash Abliterated (deepseek-v4.1-flash-abliterated), served by Experiential Labs and Refuseless, with refusal behavior removed. $0.242 input, $0.0495 cached input, $0.825 output per 1M tokens. [Model page](https://inference.dagon.so/models/deepseek-v4.1-flash-abliterated)
- **New model: GLM-5.3 Flash Abliterated** New model: GLM-5.3 Flash Abliterated (glm-5.3-flash-abliterated), served by Experiential Labs and Refuseless, with refusal behavior removed. $0.165 input, $0.066 cached input, $0.55 output per 1M tokens. [Model page](https://inference.dagon.so/models/glm-5.3-flash-abliterated)
- **Price cut: Abliterated Large v2** Abliterated Large v2 (abliterated-model-large-v2): input $3.30 to $1.375, cached input $0.33 to $0.275, output $5.50 to $2.475 per 1M tokens, markup included. [Model page](https://inference.dagon.so/models/abliterated-model-large-v2)
- **New provider for Abliterated Large v2: Experiential Labs and Refuseless** Abliterated Large v2 (abliterated-model-large-v2): now also served by Experiential Labs and Refuseless. [Model page](https://inference.dagon.so/models/abliterated-model-large-v2)

## 2026-10-26

- **New model: Abliterated Large v2** New model: Abliterated Large v2 (abliterated-model-large-v2), served by abliteration.ai, with refusal behavior removed. $3.30 input, $0.33 cached input, $5.50 output per 1M tokens. [Model page](https://inference.dagon.so/models/abliterated-model-large-v2)
- **New model: Abliterated Large Flash** New model: Abliterated Large Flash (abliterated-model-large-flash), served by abliteration.ai, with refusal behavior removed. $1.10 input, $0.11 cached input, $2.20 output per 1M tokens. [Model page](https://inference.dagon.so/models/abliterated-model-large-flash)
- **New model: Abliterated Model** New model: Abliterated Model (abliterated-model), served by abliteration.ai, with refusal behavior removed. $1.10 input, $0.11 cached input, $3.30 output per 1M tokens. [Model page](https://inference.dagon.so/models/abliterated-model)

## How this list is kept

Each site build compares the catalog it publishes with the one this list last recorded and adds an entry for every difference, dated that day. Prices are the cheapest provider's rate plus the 10% markup and move with the providers, so a change smaller than 5% is not listed until the moves add up. A removed or renamed id returns `404 model_not_found`.