[whatmodel.sh](../../README.md) / Qwen3-Coder-Next (80B-A3B) # Qwen3-Coder-Next (80B-A3B) Alibaba · released 2026-02 · 79.7B params (3B active) · apache-2.0 **score 66.8** ±1.7 · #115 of 732 · from 5 benchmark results across 2 sources Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at _{● marks Qwen3-Coder-Next (80B-A3B) among every model on that task or test, worst to best.}

task         vs the others            rank  counts  tests
agent tasks                 not ranked yet       —      4
coding       ▁▁▁▃▅▇█●▅▄▃▁     #179 of 469*    8.9%      4
math         ▁▁▃▅█▆▄█●▂▁▁      #65 of 431*     16%      1

agent tasks · 0.0% of its score

test                                result  vs the others in its run   difficulty            counts  why
SWE-bench Verified (maker)           70.6%  ▁▁▁▁▁▁▂▄●▆▁▁ beats 83/138  hard · avg 19%             —  not counted yet: no other model in the …
SWE-bench Multilingual (maker)       62.8%  ▁··▁▃▁▅●█▂▂▁ beats 24/65   hard · avg 18%             —  not counted yet: no other model in the …
SWE-bench Pro (split unspecified)…   44.3%  ▂·▁▁▁▁▂▃●▇▂▁ beats 22/58   very hard · avg 3.9%       —  not counted yet: no other model in the …
Terminal-Bench 2.0 (maker)           36.2%  ▂▂▅▅●███▆▂▃▃ beats 10/36   very hard · avg 10%        —  not counted yet: no other model in the …

coding · #179 of 469 · 8.9% of its score

Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 8B Reasoning 2512](ministral-3-8b-reasoning.md), [Phi-4-reasoning 14B](phi-4-reasoning.md) here, though they rank lower overall.

test                                result  vs the others in its run   difficulty            counts  why
HumanEval (Vector Institute, Insp…   91.5%  ▄·▄·█·●·█·▄▄ beats 5/9     easy · avg 77%          8.9%  sharp · maxed out
SWE-bench Verified (maker)           70.6%  ▁▁▁▁▁▁▂▄●▆▁▁ beats 83/138  hard · avg 19%             —  not counted yet: no other model in the …
SWE-bench Multilingual (maker)       62.8%  ▁··▁▃▁▅●█▂▂▁ beats 24/65   hard · avg 18%             —  not counted yet: no other model in the …
SWE-bench Pro (split unspecified)…   44.3%  ▂·▁▁▁▁▂▃●▇▂▁ beats 22/58   very hard · avg 3.9%       —  not counted yet: no other model in the …

math · #65 of 431 · 16% of its score

Beats [Qwen3.6 Max Preview](qwen3-6-max-preview.md), [Claude Sonnet 4.6](claude-sonnet-4-6.md), [Claude Opus 4.5](claude-opus-4-5.md) here, though they rank higher overall. Trails [INTELLECT-3](intellect-3.md), [GPT-5 mini](gpt-5-mini.md) here, though they rank lower overall.

test                                result  vs the others in its run  difficulty        counts  why
MATH (Vector Institute, Inspect E…   97.8%  ▄··██▄··▄·▄● beats 9/9    medium · avg 50%     16%  sharp · maxed out

other tests (3)

test                                result  vs the others in its run  difficulty        counts  why
GPQA Diamond (Vector Institute, I…   75.6%  ▃··▃▃▃▃·▃·▃● beats 9/9    hard · avg 40%       36%  sharp · at its level
DSEffi-Bench Pass (1,000 data-sci…   50.5%  ▅··▃▅·█·●██▃ beats 8/15   hard · avg 23%       21%  at its level
MMLU-Pro (Vector Institute, Inspe…   82.5%  ▃····▃▅·▃▅·● beats 7/9    medium · avg 58%     17%  typical

## How the score was worked out Each result implies a score on its own, after adjusting for how hard that test is and who ran it. The model's score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its *counts* column above; here it is added up by who ran the tests. [Method](../method.md)

source                                    results  share of the score
Vector Institute Evaluation Leaderboard         4   77% ▇▇▇▇▇▇▇▇▇
DSEffi-Bench: Demystifying Large Langua…        1   21% ▇▇▇
prior: pull toward the middle                        1% ▇
weighted average                                   = 66.8

_{Licences: Vector Institute Evaluation Leaderboard ([Apache-2.0](../../licenses/vector-inspect-evals-leaderboard-Apache-2.0.txt)) · DSEffi-Bench: Demystifying Large Language Models' Capability in Efficient Data Science Code Generation ([CC BY 4.0](../../licenses/paper-2608.30248v1-CC-BY-4.0.txt))}

Not counted (4)

source                  test                            score  why
Qwen/Qwen3-Coder-Next…  SWE-bench Verified              70.6%  too few models have it yet
Qwen/Qwen3-Coder-Next…  SWE-bench Multilingual          62.8%  too few models have it yet
Qwen/Qwen3-Coder-Next…  SWE-bench Pro (split unspecif…  44.3%  too few models have it yet
Qwen/Qwen3-Coder-Next…  Terminal-Bench 2.0              36.2%  too few models have it yet

## How to run it ### Through an API api (Alibaba Cloud Model Studio) · $0.60/M blended · $0.3 in / $1.5 out per 1M tokens · [alibabacloud.com](https://www.alibabacloud.com/help/en/model-studio/model-pricing) _{[price source](https://www.alibabacloud.com/help/en/model-studio/model-pricing): USD per 1M tokens, Alibaba Cloud Model Studio, Singapore (International deployment scope), qwen3-coder-next, lowest input tier (0 \< input tokens per request ≤ 32K). Requests with 32K–128K input are billed $0.5/$2.5 and 128K–256K $0.8/$4. Free quota and context-cache discounts excluded. Currency: USD. Alibaba Cloud official Model Studio pricing page, observed 2026-09-25. · [context source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/models/alibaba/qwen3-coder-next.toml): model catalogue context ceiling (not a promise by a hosted API)} ### 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 Qwen3-Coder-Next (80B-A3B) (ud-q2_k_xl, needs ~24.41 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/Qwen3-Coder-Next-GGUF/tree/main)) ```sh llama cli -hf unsloth/Qwen3-Coder-Next-GGUF:UD-Q2_K_XL ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf unsloth/Qwen3-Coder-Next-GGUF:UD-Q2_K_XL ``` ### 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 Qwen3-Coder-Next (80B-A3B) (4bit, needs ~44.8 GB, Mac only, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen3-Coder-Next-MLX-4bit/tree/03cba26036330b7553d252c1f3fb899f16cc5ea5)) ```sh lms get https://huggingface.co/lmstudio-community/Qwen3-Coder-Next-MLX-4bit -y ``` 3. Then run lms ls to see its name, and lms chat followed by that name Other versions: ud-q2_k_xl (~24.41 GB) `lms get https://huggingface.co/unsloth/Qwen3-Coder-Next-GGUF@UD-Q2_K_XL -y` · 8bit (~84.7 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Qwen3-Coder-Next-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 Qwen3-Coder-Next (80B-A3B) (Q4_K_M, needs ~48.32 GB, from [ollama.com](https://ollama.com/library/qwen3-coder-next/tags)) ```sh ollama run qwen3-coder-next ``` ### 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 Qwen3-Coder-Next (80B-A3B) (4bit, needs ~44.8 GB, Mac only, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen3-Coder-Next-MLX-4bit/tree/03cba26036330b7553d252c1f3fb899f16cc5ea5)) ```sh mlx_lm.chat --model lmstudio-community/Qwen3-Coder-Next-MLX-4bit ``` 3. Or start an OpenAI-compatible server at http://localhost:8080/v1 ```sh mlx_lm.server --model lmstudio-community/Qwen3-Coder-Next-MLX-4bit ``` Other versions: 8bit (~84.7 GB, Mac only) `mlx_lm.chat --model lmstudio-community/Qwen3-Coder-Next-MLX-8bit` Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-Coder-Next) ## Published speed measurements Reference results for the hardware, engine and settings below. These are not verified speeds of the downloads above or predictions for your machine.

hardware                      precision        speed  engine and settings
AMD Ryzen AI Max+ 395 (Stri…  IQ4_XS     61.91 tok/s  llama.cpp 1fd5f4803 (build_number…
DGX Spark (GB10), 122 GB      q4_k_m     61.97 tok/s  ollama (version not recorded) · l…

_{Each speed links to its source, which has the full settings.} [interactive version](https://whatmodel.sh/model/qwen3-coder-next) · [JSON](https://whatmodel.sh/api?format=json&all=1)