[whatmodel.sh](../../README.md) / Apriel 1.6 15B Thinker
# Apriel 1.6 15B Thinker
ServiceNow · released 2025-11 · 14.9B params · mit
**score 61.5** ±1.4 · #190 of 732 · from 7 benchmark results across 1 source · † no independently run results verified
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Apriel 1.6 15B Thinker among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆●▄▅▅▄▄▂▁ #208 of 313 32% 3
coding ▁▁▁▃▅▇█●▅▄▃▁ #186 of 469* 60% 2
vision & documents ▁▁▂▂▁▃▄●▆▇█▄ #96 of 153* 12% 2
agent tasks · #208 of 313 · 32% of its score
Beats [o1 (2024-12-17)](o1-2024-12-17.md), [o3-mini](o3-mini.md), [DeepSeek R1 (Jan '25)](deepseek-r1.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Mini](jamba-1-5-mini.md), [Mellum2 12B-A2.5B Instruct](mellum2-12b-a2-5b-instruct.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
τ²-bench Retail (maker) 66.7% █▄··●··▄·▄·▄ beats 3/6 medium · avg 36% 13% at its level
τ²-bench Airline (maker) 58.0% ▄▄······▄●▄▄ beats 4/6 hard · avg 28% 10% at its level
SWE-bench Verified (maker) 23.0% ●·█▄·······█ beats 1/6 very hard · avg 10% 9.4% sharp · at its level
coding · #186 of 469 · 60% of its score
Beats [Qwen3.5 9B](qwen3-5-9b.md), [Gemini 2.5 Flash](gemini-2-5-flash.md) here, though they rank higher overall. Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 8B Reasoning 2512](ministral-3-8b-reasoning.md), [OpenThinker 32B](openthinker-32b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Aider polyglot (225 exercises, tw… 37.7% ●▄····▄··█·▄ beats 0/6 very hard · avg 7.3% 51% sharp · at its level
SWE-bench Verified (maker) 23.0% ●·█▄·······█ beats 1/6 very hard · avg 10% 9.4% sharp · at its level
vision & documents · #96 of 153 · 12% of its score
Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-5 nano](gpt-5-nano.md) here, though they rank higher overall. Trails [Mistral Small 4 (119B-A6.5B)](mistral-small-4.md), [Qwen3-VL 32B Instruct](qwen3-vl-32b-instruct.md), [Qwen3.5 4B](qwen3-5-4b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
CharXiv (reasoning questions) (ma… 56.0% ▄·····●▄·▄·█ beats 1/5 hard · avg 31% 7.2% at its level
CharXiv descriptive questions (ma… 89.8% ▄····▄···●▄▄ beats 2/5 easy · avg 67% 4.3% noisy
other tests (1)
test result vs the others in its run difficulty counts why
Multi-IF (aggregation unspecified… 83.3% ▄·····▄●▄▄·▄ beats 3/6 easy · avg 66% 4.9% typical
source results share of the score
ServiceNow-AI/Apriel-1.6-15b-Thinker mo… 7 99% ▇▇▇▇▇▇▇▇▇▇▇▇
prior: pull toward the middle 1% ▇
weighted average = 61.5
Licences: ServiceNow-AI/Apriel-1.6-15b-Thinker model card ([MIT](../../licenses/cards-final-NOTICE.txt))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run Apriel 1.6 15B Thinker (gguf-q4_k_m, needs ~9.44 GB, from [huggingface.co/bartowski](https://huggingface.co/bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF/tree/main))
```sh
llama cli -hf bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF:Q4_K_M
```
Other versions: gguf-q8_0 (~18.38 GB) `llama cli -hf bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run Apriel 1.6 15B Thinker (gguf-q4_k_m, needs ~9.44 GB, from [huggingface.co/bartowski](https://huggingface.co/bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF/tree/main))
```sh
lms get https://huggingface.co/bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: gguf-q8_0 (~18.38 GB) `lms get https://huggingface.co/bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF@Q8_0 -y` · 4bit (~8.1 GB, Mac only) `lms get https://huggingface.co/cnfusion/Apriel-1.6-15b-Thinker-mlx-4Bit -y` · 8bit (~15.3 GB, Mac only) `lms get https://huggingface.co/cnfusion/Apriel-1.6-15b-Thinker-mlx-8Bit -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run Apriel 1.6 15B Thinker (gguf-q4_k_m, needs ~9.44 GB, from [huggingface.co/bartowski](https://huggingface.co/bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF/tree/main))
```sh
ollama run hf.co/bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF:Q4_K_M
```
Other versions: gguf-q8_0 (~18.38 GB) `ollama run hf.co/bartowski/ServiceNow-AI_Apriel-1.6-15b-Thinker-GGUF:Q8_0`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run Apriel 1.6 15B Thinker (4bit, needs ~8.1 GB, Mac only, from [huggingface.co/cnfusion](https://huggingface.co/cnfusion/Apriel-1.6-15b-Thinker-mlx-4Bit/tree/8cd9214f84bcbf19a7fa8994b065c6a0c6d6ca8a))
```sh
mlx_lm.chat --model cnfusion/Apriel-1.6-15b-Thinker-mlx-4Bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model cnfusion/Apriel-1.6-15b-Thinker-mlx-4Bit
```
Other versions: 8bit (~15.3 GB, Mac only) `mlx_lm.chat --model cnfusion/Apriel-1.6-15b-Thinker-mlx-8Bit`
Model repository and documentation: [huggingface.co/ServiceNow-AI](https://huggingface.co/ServiceNow-AI/Apriel-1.6-15b-Thinker)
[interactive version](https://whatmodel.sh/model/apriel-1-6-15b-thinker) · [JSON](https://whatmodel.sh/api?format=json&all=1)