[whatmodel.sh](../../README.md) / Ornith 1.5 35B-A3B # Ornith 1.5 35B-A3B DeepReinforce · released 2026-08 · 35.95B params (3B active) · mit **score 69.2** ±1.4 · #86 of 732 · from 12 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 Ornith 1.5 35B-A3B among every model on that task or test, worst to best. 
task         vs the others        rank  counts  tests
agent tasks  ▁▁▃▆█▄▅▅●▄▂▁   #78 of 313     68%      9
coding       ▁▁▁▃▅▇█▇▅●▃▁   #66 of 469     32%      3
web search   █··········●      #1 of 2     16%      3
agent tasks · #78 of 313 · 68% of its score Beats [GPT-5](gpt-5.md), [Gemini 3 Flash (Preview)](gemini-3-flash-preview.md), [Grok 4.20 (0309 reasoning)](grok-4-20.md) here, though they rank higher overall. Trails [MiMo-V2.5 (310B-A15B)](mimo-v2-5.md), [MiMo-V2.5-Pro (1T-A42B)](mimo-v2-5-pro.md), [MiniMax-M3](minimax-m3.md) here, though they rank lower overall. 
test                                result  vs the others in its run    difficulty           counts  why
SWE-bench Pro (split unspecified)…   59.6%  ▂·▁▁▁▁▂▃█●▂▁ beats 43/58    very hard · avg 10%     13%  sharp · at its level · small run
SWE-bench Verified (maker)           79.0%  ▁▁▁▁▁▁▂▄█●▁▁ beats 129/138  hard · avg 26%          12%  sharp · small run
Terminal-Bench 2.1 (maker)           67.8%  ▁·▁·▁▁▄▃▆●▅█ beats 37/63    hard · avg 20%          10%  sharp · at its level · small run
SWE-bench Multilingual (maker)       71.4%  ▁··▁▃▁▅█●▂▂▁ beats 40/65    hard · avg 29%         7.5%  at its level · small run
MCP-Atlas (public set) (maker)       70.2%  ▂·▁▂▂·▁▃█●▆▂ beats 26/39    hard · avg 30%         6.2%  at its level · small run
BrowseComp (maker)                   67.6%  ▂·▃▁▂▂▆▅●▂▂▄ beats 32/46    hard · avg 27%         6.0%  at its level · small run
Humanity's Last Exam with tools (…   33.4%  ▁·▁▂▂▁●▄▇█▄▅ beats 16/62    very hard · avg 11%    5.8%  at its level · small run
WideSearch (maker)                   67.8%  █·▂●··▂···▂▂ beats 6/10     medium · avg 37%       4.2%  at its level · small run
Toolathlon-Verified (maker)          48.7%  ▂·▂▂●▃···▂██ beats 4/18     hard · avg 27%         3.4%  blunt · at its level · small run
coding · #66 of 469 · 32% of its score Beats [Gemini 3.1 Pro (Preview)](gemini-3-1-pro.md), [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-5.2](gpt-5-2.md) here, though they rank higher overall. Trails [MiMo-V2.5-Pro (1T-A42B)](mimo-v2-5-pro.md), [MiniMax-M3](minimax-m3.md) here, though they rank lower overall. 
test                                result  vs the others in its run    difficulty           counts  why
SWE-bench Pro (split unspecified)…   59.6%  ▂·▁▁▁▁▂▃█●▂▁ beats 43/58    very hard · avg 10%     13%  sharp · at its level · small run
SWE-bench Verified (maker)           79.0%  ▁▁▁▁▁▁▂▄█●▁▁ beats 129/138  hard · avg 26%          12%  sharp · small run
SWE-bench Multilingual (maker)       71.4%  ▁··▁▃▁▅█●▂▂▁ beats 40/65    hard · avg 29%         7.5%  at its level · small run
web search · #1 of 2 · 16% of its score 
test                                result  vs the others in its run  difficulty           counts  why
BrowseComp (maker)                   67.6%  ▂·▃▁▂▂▆▅●▂▂▄ beats 32/46  hard · avg 27%         6.0%  at its level · small run
Humanity's Last Exam with tools (…   33.4%  ▁·▁▂▂▁●▄▇█▄▅ beats 16/62  very hard · avg 11%    5.8%  at its level · small run
WideSearch (maker)                   67.8%  █·▂●··▂···▂▂ beats 6/10   medium · avg 37%       4.2%  at its level · small run
other tests (3) 
test                          result  vs the others in its run    difficulty            counts  why
GPQA Diamond (maker)           89.2%  ▁▁▁▁▂▄▅▅██●▄ beats 266/322  hard · avg 45%           16%  sharp · small run
Humanity's Last Exam (maker)   25.6%  ▁▁▁▄▆▆▇●█▆▁▃ beats 65/111   very hard · avg 4.8%    8.7%  at its level · small run
NL2Repo (maker)                46.2%  ▂▂▂·▃▃▅█●▂▅▃ beats 15/22    very hard · avg 12%     6.3%  at its level · small run
## 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
ornith-ai/Ornith-1.5-35B-A3B model card…       12   99% ▇▇▇▇▇▇▇▇▇▇▇▇
prior: pull toward the middle                        1% ▇
weighted average                                   = 69.2
Licences: ornith-ai/Ornith-1.5-35B-A3B model card ([MIT](https://opensource.org/licenses/MIT)) ## 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 Ornith 1.5 35B-A3B (Q4_K_M, needs ~21.7 GB, from [huggingface.co/ornith-ai](https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-GGUF/tree/12393612fd4f730ff5aadc23e9b8f9648aa49ceb)) ```sh llama cli -hf ornith-ai/Ornith-1.5-35B-A3B-GGUF:Q4_K_M ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf ornith-ai/Ornith-1.5-35B-A3B-GGUF:Q4_K_M ``` Other versions: Q8_0 (~37.8 GB) `llama cli -hf ornith-ai/Ornith-1.5-35B-A3B-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 Ornith 1.5 35B-A3B (Q4_K_M, needs ~21.7 GB, from [huggingface.co/ornith-ai](https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-GGUF/tree/12393612fd4f730ff5aadc23e9b8f9648aa49ceb)) ```sh lms get https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-GGUF@Q4_K_M -y ``` 3. Then run lms ls to see its name, and lms chat followed by that name Other versions: Q8_0 (~37.8 GB) `lms get https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-GGUF@Q8_0 -y` · 4bit (~19.5 GB, Mac only) `lms get https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-MLX-4bit -y` · 8bit (~36.8 GB, Mac only) `lms get https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-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 Ornith 1.5 35B-A3B (Q4_K_M, needs ~23 GB, from [ollama.com](https://ollama.com/library/ornith-1.5/tags)) ```sh ollama run ornith-1.5:35b ``` ### 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 Ornith 1.5 35B-A3B (4bit, needs ~19.5 GB, Mac only, from [huggingface.co/ornith-ai](https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-MLX-4bit/tree/19504d912fa8fc7622bf6b1de3db5d5d890b1f02)) ```sh mlx_lm.chat --model ornith-ai/Ornith-1.5-35B-A3B-MLX-4bit ``` 3. Or start an OpenAI-compatible server at http://localhost:8080/v1 ```sh mlx_lm.server --model ornith-ai/Ornith-1.5-35B-A3B-MLX-4bit ``` Other versions: 8bit (~36.8 GB, Mac only) `mlx_lm.chat --model ornith-ai/Ornith-1.5-35B-A3B-MLX-8bit` Model repository and documentation: [huggingface.co/ornith-ai](https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B) ## 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
DGX Spark (GB10), 122 GB  q4_k_m     79.87 tok/s  ollama (version not recorded) · l…
Each speed links to its source, which has the full settings. [interactive version](https://whatmodel.sh/model/ornith-1-5-35b-a3b) · [JSON](https://whatmodel.sh/api?format=json&all=1)