Quick start

ollama run nous-hermes2-mixtral:latest

Runnable variants

Memory is a rough file-size estimate; context length, runtime and GPU offload change it.

12 of 18 matching variants shown.18 of 18 variants shown.

Tag	Type	Inputs	Digest	Download	Parameters	Quantization	Context	Est. memory	Run
8x7b	Typelocal	Inputstext	Digestda8c35896cff	Download26 GB	Parameters8x7b	QuantizationUnknown	Context32K	Est. memory≈ 30.5 GB	`ollama run nous-hermes2-mixtral:8x7b`
8x7b-dpo-fp16	Typelocal	Inputstext	Digest42c40087888d	Download93 GB	Parameters8x7b	QuantizationFP16	Context32K	Est. memory≈ 105 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-fp16`
8x7b-dpo-q2_K	Typelocal	Inputstext	Digestdef75f44e678	Download17 GB	Parameters8x7b	QuantizationQ2_K	Context32K	Est. memory≈ 20 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q2_K`
8x7b-dpo-q3_K_L	Typelocal	Inputstext	Digestd3f241b3fa43	Download24 GB	Parameters8x7b	QuantizationQ3_K_L	Context32K	Est. memory≈ 28 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q3_K_L`
8x7b-dpo-q3_K_M	Typelocal	Inputstext	Digeste401c6cff6ee	Download23 GB	Parameters8x7b	QuantizationQ3_K_M	Context32K	Est. memory≈ 27 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q3_K_M`
8x7b-dpo-q3_K_S	Typelocal	Inputstext	Digestb83102712906	Download20 GB	Parameters8x7b	QuantizationQ3_K_S	Context32K	Est. memory≈ 23.5 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q3_K_S`
8x7b-dpo-q4_0	Typelocal	Inputstext	Digestda8c35896cff	Download26 GB	Parameters8x7b	QuantizationQ4_0	Context32K	Est. memory≈ 30.5 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q4_0`
8x7b-dpo-q4_1	Typelocal	Inputstext	Digest6b8631272ae5	Download29 GB	Parameters8x7b	QuantizationQ4_1	Context32K	Est. memory≈ 33.5 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q4_1`
8x7b-dpo-q4_K_M	Typelocal	Inputstext	Digest5ce16766078d	Download28 GB	Parameters8x7b	QuantizationQ4_K_M	Context32K	Est. memory≈ 32.5 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q4_K_M`
8x7b-dpo-q4_K_S	Typelocal	Inputstext	Digest72597fc38d4a	Download27 GB	Parameters8x7b	QuantizationQ4_K_S	Context32K	Est. memory≈ 31.5 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q4_K_S`
8x7b-dpo-q5_0	Typelocal	Inputstext	Digest2d30ec45b54c	Download32 GB	Parameters8x7b	QuantizationQ5_0	Context32K	Est. memory≈ 37 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q5_0`
8x7b-dpo-q5_1	Typelocal	Inputstext	Digestd37f3798cec9	Download35 GB	Parameters8x7b	QuantizationQ5_1	Context32K	Est. memory≈ 40.5 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q5_1`
8x7b-dpo-q5_K_M	Typelocal	Inputstext	Digestc95f2fb1bbb0	Download33 GB	Parameters8x7b	QuantizationQ5_K_M	Context32K	Est. memory≈ 38 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q5_K_M`
8x7b-dpo-q5_K_S	Typelocal	Inputstext	Digest78bf09c60f45	Download32 GB	Parameters8x7b	QuantizationQ5_K_S	Context32K	Est. memory≈ 37 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q5_K_S`
8x7b-dpo-q6_K	Typelocal	Inputstext	Digestaf9b1e15b356	Download38 GB	Parameters8x7b	QuantizationQ6_K	Context32K	Est. memory≈ 43.5 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q6_K`
8x7b-dpo-q8_0	Typelocal	Inputstext	Digesta6f9a3257097	Download50 GB	Parameters8x7b	QuantizationQ8_0	Context32K	Est. memory≈ 57 GB	`ollama run nous-hermes2-mixtral:8x7b-dpo-q8_0`
dpo	Typelocal	Inputstext	Digestda8c35896cff	Download26 GB	ParametersUnknown	QuantizationUnknown	Context32K	Est. memory≈ 30.5 GB	`ollama run nous-hermes2-mixtral:dpo`
latestdefault	Typelocal	Inputstext	Digestda8c35896cff	Download26 GB	ParametersUnknown	QuantizationUnknown	Context32K	Est. memory≈ 30.5 GB	`ollama run nous-hermes2-mixtral:latest`

Related models

qwen3.6local

Qwen3.6 delivers substantial upgrades in agentic coding and thinking preservation than previous Qwen models.

mixtrallocal

A set of Mixture of Experts (MoE) model with open weights by Mistral AI in 8x7b and 8x22b parameter sizes.

dolphin-mixtrallocal

Uncensored, 8x7b and 8x22b fine-tuned models based on the Mixtral mixture of experts models that excels at coding tasks. Created by Eric Hartford.

muse-glimmerlocal

Meta's latest open model built for always-on local agents. 30B parameters, licensed under Apache 2.0 and runs on a single GPU — tuned for tool use, long tasks, and failure recovery.