Mistral relicenses Large 3 and Small 4 under Apache 2.0, reversing its earlier restrictive turn
Both Mistral Large 3 and Mistral Small 4 now ship under Apache 2.0 — a significant reversal for a lab that had drifted toward restrictive licensing on its larger models. The move puts Mistral's flagship weights back in the genuinely-open column alongside Qwen's Apache 2.0 and DeepSeek's MIT, and reshapes what a European enterprise can deploy without licence friction.
The licence is the substance here, not the model number. Mistral had increasingly reserved its most capable weights under more restrictive terms, and moving Large 3 and Small 4 to Apache 2.0 reverses that — Apache 2.0 permits commercial use, modification, and redistribution with minimal obligation, which is the difference between a model an enterprise can build a product on and one it can only evaluate.
The competitive pressure is Chinese and permissive. With Qwen under Apache 2.0 and DeepSeek under MIT setting the expectation that frontier-adjacent open weights come with liberal licences, a European lab holding its best models under restrictive terms was ceding the deployable-open-weights market. Relicensing is Mistral choosing to compete on the terms the market has settled on.
For buyers the effect is a wider menu of genuinely-open flagships with clear provenance. A Western enterprise wary of the provenance of the leading Chinese open models now has a European Apache-2.0 option at the flagship tier — which matters more, not less, as the EU's training-data-summary duties become enforceable and licence terms become a compliance input.
Thunder Compute — Best open source LLMs (August 2026) → · Hugging Face — Best open-source LLM models in 2026: coding, local, agentic, benchmarks, license →