Mistral Medium 3.5 Review: A 128B Open-Weight Model With a Coding Agent That Opens PRs For You
Mistral's Medium 3.5 is a 128B open-weight model that runs on 4 GPUs, scores 77.6% on SWE-Bench Verified, and ships with Vibe — a cloud coding agent t
Search for a command to run...
Articles tagged with #open-source-ai
Mistral's Medium 3.5 is a 128B open-weight model that runs on 4 GPUs, scores 77.6% on SWE-Bench Verified, and ships with Vibe — a cloud coding agent t
Moonshot AI's Kimi K2.6 is a 1-trillion parameter open-weight model that outscores GPT-5.4 on SWE-Bench Pro, runs 300-agent swarms, and costs $0.60 pe
Alibaba's Qwen3.6-35B-A3B packs frontier-class coding performance into a model that fits on a single 24GB GPU. Here's what it actually means for local
Llama 4 Scout has a 10M token context window, multimodal by default, and runs on 12GB VRAM. But Meta's Llama 4 Community License isn't Apache 2.0 -- a
Can you use Gemma 4 commercially without paying Google? Yes -- and that's the most important thing about this model. Here's what Apache 2.0 actually m
Flux 1.1 Pro and Stable Diffusion SDXL/SD3.5 are the two main open-source AI image generation architectures. Here's which one actually wins for real-w