Supported Models

All open-weight models supported, plus any custom or self-trained GGUF models.

AlphaChat downloads and optimizes models automatically. You can also load your own models — any GGUF file works.

alphachat run <model-id>

9 models

DeepSeek V4 Flashdeepseek-v4-flash
highMoE

params

284B (13B active)

Context

128K

Architecture

MoE

Run

alphachat run deepseek-v4-flash

Qwen 3.6 35Bqwen-3.6-35b
highMoE

params

35B (3B active)

Context

128K

Architecture

MoE

Run

alphachat run qwen-3.6-35b

Mixtral 8x7Bmixtral-8x7b
mediumMoE

params

46.7B (12.9B active)

Context

32K

Architecture

MoE

Run

alphachat run mixtral-8x7b

Qwen3.5 122Bqwen3.5-122b
mediumMoE

params

122B (8B active)

Context

128K

Architecture

MoE

Run

alphachat run qwen3.5-122b

Qwen 3.5 27Bqwen3.5-27b
lowDense

params

27B

Context

128K

Architecture

Dense

Run

alphachat run qwen3.5-27b

Qwen 3.5 9Bqwen3.5-9b
lowDense

params

9B

Context

32K

Architecture

Dense

Run

alphachat run qwen3.5-9b

Qwen 3.5 4Bqwen3.5-4b
lowDense

params

4B

Context

32K

Architecture

Dense

Run

alphachat run qwen3.5-4b

Gemma4 E4Bgemma4-e4b
lowDense

params

4B

Context

32K

Architecture

Dense

Run

alphachat run gemma4-e4b

DeepSeek-R1 14Bdeepseek-r1-14b
lowDense

params

14B

Context

128K

Architecture

Dense

Run

alphachat run deepseek-r1-14b