Ornith-1.5 is a new family of open-source large language models comprising 9B Dense, 35B MoE, and 397B MoE variants trained using self-improving strategies. The models achieve state-of-the-art performance among open-source models of comparable size and deliver results comparable to Claude Opus 4.8 across reasoning, agentic, and coding tasks.
| Benchmark | Model | Score |
|---|---|---|
| Terminal-Bench 2 | Ornith-1.5 | 86.1% |
| SWE-bench Verified | Ornith-1.5 | 86% |
| SWE-bench Multilingual | Ornith-1.5 | 79.6% |
| SWE-bench Pro | Ornith-1.5 | 65.1% |
| DeepSWE | Ornith-1.5 | 56% |
| Humanity's Last Exam | Ornith-1.5 | 44.6% |
Alibaba's Qwen team has announced Qwen 3.8 Max, a new 2.4T-parameter flagship model focused on coding, long-horizon agentic work, and multimodal reasoning. The company confirmed that open-weight versions of both Qwen 3.8 Max and the smaller Qwen 3.8-27B will be released next week.
Poolside has released Laguna S 2.1, a 118B-parameter Mixture-of-Experts model designed for agentic coding tasks. The weights are available on Hugging Face under an OpenMDW-1.1 license and can run on a single NVIDIA DGX Spark.
Alibaba announced the open-weight release of Qwen3.8, a 2.4-trillion-parameter model, while also open-sourcing its Zhenwu AI chip software stack to reduce reliance on Nvidia's CUDA ecosystem.
Moonshot's Kimi K3 launch has triggered a broad reassessment of how close Chinese open-weight models are to the AI frontier, with community reaction ranging from congratulations to concerns that it pressures US labs to ship faster.
SpaceXAI has open-sourced Grok Build, the terminal-based AI coding agent powering its grok CLI, under the Apache 2.0 license. The release includes the source code for the agent harness, terminal user interface (TUI), and tool layer.
We use cookies to measure traffic and improve the site. You can accept or decline analytics cookies. Privacy policy