[whatmodel.sh](../../README.md) / Ling 3.0 Flash (124B-A5B) # Ling 3.0 Flash (124B-A5B) inclusionAI (Ant Group) · released 2026-08 · 124B params (5.1B active) · mit **score 69.9** ±1.5 · #80 of 732 · from 13 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 Ling 3.0 Flash (124B-A5B) among every model on that task or test, worst to best. 
task         vs the others     rank  counts  tests
agent tasks  ······●·····   #1 of 1     27%      3
coding       ······●·····   #1 of 1     28%      3
math         ······●·····   #1 of 1     31%      3
agent tasks · #1 of 1 · 27% of its score 
test                                result  vs the others in its run    difficulty            counts  why
SWE-bench Pro (split unspecified)…   56.6%  ▂·▁▁▁▁▂▃●▇▂▁ beats 37/58    very hard · avg 9.1%     15%  sharp · at its level · small run
SWE-bench Multilingual (maker)       72.4%  ▁··▁▃▁▅█●▂▂▁ beats 42/65    hard · avg 24%          7.5%  at its level · small run
BFCL V4 (December 2025 archive) (…   73.0%  ▁▁▁▂▆▆▅▆█▂▄● beats 112/114  medium · avg 38%        4.7%  at its level · small run
coding · #1 of 1 · 28% of its score 
test                                result  vs the others in its run  difficulty            counts  why
SWE-bench Pro (split unspecified)…   56.6%  ▂·▁▁▁▁▂▃●▇▂▁ beats 37/58  very hard · avg 9.1%     15%  sharp · at its level · small run
SWE-bench Multilingual (maker)       72.4%  ▁··▁▃▁▅█●▂▂▁ beats 42/65  hard · avg 24%          7.5%  at its level · small run
LiveCodeBench 2024-08–2025-05 (ve…   82.8%  ▃···█··▅··●▃ beats 6/7    medium · avg 51%        5.3%  small run
math · #1 of 1 · 31% of its score 
test                        result  vs the others in its run  difficulty        counts  why
AIME 2026 (maker)            93.2%  ▁▁▁▂▂▂▆●▆▂·▁ beats 36/52  medium · avg 60%     11%  sharp · maxed out · small run
IMO-AnswerBench (maker)      83.7%  ▁▁·▁▁▁▄▆█●▂▃ beats 32/44  medium · avg 40%     10%  small run
HMMT February 2026 (maker)   87.0%  ▃▃▂▅·████●▃▅ beats 29/35  medium · avg 42%     10%  sharp · small run
other tests (8) 
test                                result  vs the others in its run   difficulty            counts  why
Humanity's Last Exam (maker)         22.7%  ▁▁▁▄▆▆▇●█▆▁▃ beats 57/111  very hard · avg 4.0%     10%  small run
SkillsBench 87-task subset (maker)   44.8%  ▄··▄▄▄···●·█ beats 4/7     very hard · avg 9.1%    7.3%  at its level
WideSearch corrected dataset (mak…   73.6%  ▃······▃▃▃●▃ beats 4/7     hard · avg 22%          6.6%  at its level · noisy
SysBench (aggregation unspecified…   93.6%  █··█····█··● beats 3/3     medium · avg 64%        4.3%  small run
Multi-IF (aggregation unspecified…   87.7%  ▁▁▁▁▁▁▃▄█▄▁● beats 40/41   easy · avg 66%          3.9%  small run
IFBench (aggregation unspecified)…   74.5%  ▂▆▂▄▄▄▂▄▄▆●▇ beats 64/86   hard · avg 33%          3.3%  at its level · noisy · small run
MCP-Atlas v1 public500 (maker)       65.5%  █··········● beats 1/1                                —  not counted yet: no other model in the …
BrowseComp single-agent with cont…   72.2%  ●··········█ beats 0/1                                —  not counted yet: no other model in the …
## 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
inclusionAI/Ling-3.0-flash model card (…       13   99% ▇▇▇▇▇▇▇▇▇▇▇▇
prior: pull toward the middle                        1% ▇
weighted average                                   = 69.9
Licences: inclusionAI/Ling-3.0-flash model card ([MIT](../../licenses/cards-final-NOTICE.txt))
Not counted (2) 
source                  test                            score  why
inclusionAI/Ling-3.0-…  MCP-Atlas v1 public500          65.5%  too few models have it yet
inclusionAI/Ling-3.0-…  BrowseComp single-agent with …  72.2%  too few models have it yet
## How to run it ### Through an API api (DeepInfra) · $0.090/M blended · $0.06 in / $0.18 out per 1M tokens · [deepinfra.com](https://deepinfra.com/inclusionAI/Ling-3.0-flash) [price source](https://github.com/BerriAI/litellm/blob/c19ce71bcdad234ca87a7af10cee45375ce811a8/model_prices_and_context_window.json): USD per 1M text tokens. DeepInfra standard synchronous API; model inclusionAI/Ling-3.0-flash. Uncached input; output includes generated reasoning where billed. Provider input limit 131,072 tokens; quantization not specified in the provider record. Batch, cache storage, tools and other modalities excluded. · [context source](https://github.com/BerriAI/litellm/blob/c19ce71bcdad234ca87a7af10cee45375ce811a8/model_prices_and_context_window.json): DeepInfra API maximum input tokens from exact catalogue record deepinfra/inclusionAI/Ling-3.0-flash; not the legacy max\_tokens field. ### 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 Ling 3.0 Flash (124B-A5B) (Q4_K_M, needs ~74.9 GB, from [huggingface.co/inclusionAI](https://huggingface.co/inclusionAI/Ling-3.0-flash-GGUF/tree/main)) ```sh llama cli -hf inclusionAI/Ling-3.0-flash-GGUF:Q4_K_M ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf inclusionAI/Ling-3.0-flash-GGUF:Q4_K_M ``` ### 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 Ling 3.0 Flash (124B-A5B) (Q4_K_M, needs ~74.9 GB, from [huggingface.co/inclusionAI](https://huggingface.co/inclusionAI/Ling-3.0-flash-GGUF/tree/main)) ```sh lms get https://huggingface.co/inclusionAI/Ling-3.0-flash-GGUF@Q4_K_M -y ``` 3. Then run lms ls to see its name, and lms chat followed by that name Other versions: 4bit (~70 GB, Mac only) `lms get https://huggingface.co/TensorFold/Ling-3.0-flash-MLX-4bit -y` · 8bit (~132.2 GB, Mac only) `lms get https://huggingface.co/TensorFold/Ling-3.0-flash-MLX-8bit -y` ### 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 Ling 3.0 Flash (124B-A5B) (4bit, needs ~70 GB, Mac only, from [huggingface.co/TensorFold](https://huggingface.co/TensorFold/Ling-3.0-flash-MLX-4bit/tree/cae62774492b589eea4aab2d6d36fb167a67d90c)) ```sh mlx_lm.chat --model TensorFold/Ling-3.0-flash-MLX-4bit ``` 3. Or start an OpenAI-compatible server at http://localhost:8080/v1 ```sh mlx_lm.server --model TensorFold/Ling-3.0-flash-MLX-4bit ``` Other versions: 8bit (~132.2 GB, Mac only) `mlx_lm.chat --model TensorFold/Ling-3.0-flash-MLX-8bit` Model repository and documentation: [huggingface.co/inclusionAI](https://huggingface.co/inclusionAI/Ling-3.0-flash) [interactive version](https://whatmodel.sh/model/ling-3-0-flash) · [JSON](https://whatmodel.sh/api?format=json&all=1)