Open-weight models
Models whose trained weights are downloadable (run them yourself) — distinct from “fully open source” (weights + training code + data, which is rare). The defining 2026 movement in the llm-provider landscape open-source-llms-2026.
State of play (2026)
Open-weight is now “good enough for serious production,” not just a cheap alternative. Leaders: deepseek V4 Pro/R1, Meta Llama 4, Alibaba Qwen3, google gemma-4, Moonshot Kimi K2.6, Z.ai GLM-5.1, Mistral Large 3/Small 4, Microsoft Phi-4.
Three structural trends
- MoE dominance — almost every flagship is a sparse Mixture-of-Experts: large total params, small active set (Gemma 4 25.2B/3.8B; Llama 4 Maverick 400B/17B; DeepSeek R1 671B/37B). Gives frontier capability at practical inference cost — the architectural lever behind cheap serving (cf. llm-inference, kv-cache).
- Context explosion — 256K → 1M → 10M (Llama 4 Scout).
- Licensing clarity — Apache-2.0 won the permissive race (Qwen3, Gemma 4, Mistral Large 3); MIT for Phi-4 and DeepSeek R1. Meta’s Llama “Community” license is more restrictive; “read the terms” applies (Kimi’s Modified MIT).
- Multimodality reaches the open/local tier — open weights are no longer text-only: gemma-4 12B ships encoder-free vision + native audio (raw audio projected straight into the token space), runnable on 16GB VRAM/unified memory gemma-4-12b-announcement. Memory footprint, not just licensing, is becoming the open-weight battleground.
Why it matters
Open weights enable local/private deployment (Ollama, vLLM, llama.cpp) and break the pricing floor — the competitive pressure that makes deepseek‘s low llm-api-pricing possible and forces proprietary labs to justify their premium on llm-benchmarks.
Related
open-source-llms-2026 · deepseek · llm-provider · llm-benchmarks · llm-inference