Ornith 1.5
An open-source Mixture-of-Experts (MoE) large language model released by Deep Green Force under the MIT license. It features 35 billion total parameters with 256 experts, of which 8 are activated per forward pass, and is designed to run on consumer-grade graphics cards.
Timeline (last 90 days)
With reasoning mode turned off, Qwen3.8-27B detected 9 of 20 planted bugs, and Ornith 1.5 detected 8, a difference attributable to statistical noise.
Qwen3.8-27B occupies 16.1GB of memory after loading, and Ornith 1.5 occupies 19.8GB of memory.
After enabling reasoning mode, Qwen3.8-27B's bug detection rate dropped to 6/20, and 12 of 40 answers could not be parsed as JSON because they exhausted the 12,000-token output limit; Ornith 1.5's detection rate remained 8/20, with 13/40 answers unparseable, but it took about half the time of Qwen.
With an 800 token input prompt, Qwen3.8-27B's time to first token is 3.46 seconds and Ornith 1.5's is 0.49 seconds, the latter's time to first token being about one seventh of the former's.
Qwen3.8-27B是稠密模型，全部270亿参数参与每个token的计算，依靠MTP头预测后续token并在单次前向中验证；Ornith 1.5采用MoE架构，每token仅激活约30亿参数，配合DFlash由小型草稿模型提议token块再由主模型校验。
Qwen3.8-27B's output speed is 55.1 tokens/second, and Ornith 1.5's is 90.4 tokens/second.
Ornith 1.5 adopts a three-stage reinforcement learning architecture, in which the first-stage task generator autonomously analyzes codebases to generate training tasks and receives rewards based on correctness, difficulty, and novelty.
Deep Reinforce's Ornith 1.5 35B was released on August 19, 2026.
Ornith 1.5 35B is a Mixture-of-Experts (MoE) model with 35 billion total parameters, activating approximately 3 billion per token, containing 256 independent experts internally, with a routing mechanism that selects 8 of them to process each token.
Ornith 1.5 uses the MIT license and is one member of a three-model family that includes a 9B dense version, a 35B MoE version, and a 397B flagship version.
15 more timeline events
All Facts (18)
In a test of 30 Three.js One-Shot programming problems, Qwen 3.8 passed 24 and Ornith 1.5 passed 15, both running under Q6 quantization.
60%UnverifiedQwen3.8-27B是稠密模型，全部270亿参数参与每个token的计算，依靠MTP头预测后续token并在单次前向中验证；Ornith 1.5采用MoE架构，每token仅激活约30亿参数，配合DFlash由小型草稿模型提议token块再由主模型校验。
50%UnverifiedOrnith 1.5 adopts a three-stage reinforcement learning architecture, in which the first-stage task generator autonomously analyzes codebases to generate training tasks and receives rewards based on correctness, difficulty, and novelty.
50%UnverifiedOrnith 1.5 35B is a Mixture-of-Experts (MoE) model with 35 billion total parameters, activating approximately 3 billion per token, containing 256 independent experts internally, with a routing mechanism that selects 8 of them to process each token.
50%UnverifiedOrnith 1.5 uses the MIT license and is one member of a three-model family that includes a 9B dense version, a 35B MoE version, and a 397B flagship version.
50%UnverifiedIn a 3D game building test on an Apple M3 Ultra with 512GB unified memory, Qwen 3.8 completed it in one shot, while Ornith 1.5 required 6 separate iterations to barely approach the same level.
50%UnverifiedIn an ultra-long context test, Ornith 1.5 ran at around 75 tokens/second early on and dropped to approximately 45 tokens/second after exceeding 1.6 million tokens, while Qwen 3.8 dropped to approximately 5 tokens/second under the same load.
50%UnverifiedBoth models are supported by llama.cpp, Ollama, and LM Studio, and both support MTP multi-token prediction speculative decoding acceleration.
50%UnverifiedThe Ornith 1.5 family includes a 9B compact version and a 397B flagship version, while Qwen has a Max version with 2.4 trillion parameters available via API.
50%UnverifiedOrnith 1.5 uses the MIT license and can be freely modified and used commercially.
50%UnverifiedOrnith 1.5 is a Mixture of Experts (MoE) model with 256 experts internally, activating only 8 experts per token, with approximately 3B parameters actually activated.
50%UnverifiedThe routing in a MoE model dynamically selects experts on a per-token basis, so all 35B parameters must reside in VRAM simultaneously—computation is cheap but VRAM usage is not reduced.
50%UnverifiedIn a task to generate a 3D game from scratch, Qwen 3.8 completed it with a single prompt and zero guidance, while Ornith 1.5 required six iterations.
50%UnverifiedOrnith 1.5 requires approximately 23 GB of VRAM for the recommended Q4_K_M quantization with the full package, and it is advised to prepare 27 GB or more; Qwen 3.8 at the same Q4_K_M tier requires only 16 to 19 GB.
50%UnverifiedThe Ornith team simultaneously provides GGUF quantized versions for each model.
50%UnverifiedDeep Reinforce's Ornith 1.5 35B was released on August 19, 2026.
50%UnverifiedDeep Green Force released the open-source model Ornith 1.5 on August 19, with a parameter scale of 35B.
50%UnverifiedThe Ornith AI team released three open-source models in the Ornith 1.5 series, covering three parameter scales: 9B, 35B-A3B, and 397B.
50%