[Late catch] Mistral Large 4 preview: a 1T-parameter model with open weights on the way
What is it?
Mistral's largest model yet is a mixture-of-experts (MoE) model, an architecture that only activates a subset of expert layers per token. It has 1 trillion parameters in total, 52 billion active. It understands images and documents alongside text and supports more than 160 languages. For now it's a public API preview in Mistral Studio, with weights promised by the end of the month.
Why does it matter?
At $1.36 per million input tokens and $4.18 per million output, it costs far less than the top closed models, and once weights ship, on-prem and private-cloud deployment become possible. Mistral is focusing on agent and security work, citing 61.7% on DeepSWE v1.1Vendor claim and 93% on CybenchVendor claim. That matters most to European, public-sector and financial teams that care about data sovereignty.
Maturity
PreviewIt's still an API preview, and the weights and license terms aren't out yet. Nearly all benchmarks come from Mistral itself, so wait for independent results after the weights land.
How to track it
- Weights and license terms: watch Mistral's news page for the release due by the end of this month.
- Independent results: once the weights land, compare Mistral's own DeepSWE and Cybench numbers with outside benchmarks.
- General availability and whether pricing holds after the preview: follow Mistral's news page for the Studio API launch.