Monthly signal
The standout launches, model drops, and platform moves worth tracking.
Mistral AI · Mistral Small
Mistral AI released Mistral Small 4 on March 16, 2026.
Mistral Small 4 is the Mistral AI model released on March 16, 2026, a hybrid model that unifies the reasoning of Magistral, the multimodality of Pixtral and the agentic coding of Devstral in one model for chat, coding, agentic tasks and complex reasoning. It is a mixture-of-experts model with 128 experts (4 active per token), 119B total parameters and 6B active per token, with a 256k context window and text and image input, released under Apache 2.0. Reasoning is configurable with reasoning_effort, from none for fast replies to high for step-by-step reasoning. API pricing is $0.15 per million input tokens and $0.60 per million output tokens.
mistral-small-2603Compared with Mistral Small 3.x, Mistral reports a 40% reduction in end-to-end completion time in a latency-optimised setup and 3x more requests per second in a throughput-optimised setup.
It also folds reasoning, vision and agentic coding, previously separate Mistral models, into one model with adjustable reasoning effort.
New to model names and version numbers? Read how AI labs version their models.
The launch and every later Mistral AI release that mentions Mistral Small 4, oldest first.
Mistral released Small 4, a hybrid multimodal model that unifies instruct, reasoning, and coding with a 256k context window.
Source: Mistral Docs ChangelogWe released Mistral Small 4 ( mistral-small-2603 ), a hybrid model unifying instruct, reasoning, and coding in a single multimodal model with a 256k context window.
Source: Mistral Docs ChangelogMistral AI released Mistral Small 4 on March 16, 2026. The launch entry below links to the official announcement.
Mistral Small 4 is available on Mistral API, Mistral AI Studio, Hugging Face and NVIDIA NIM. Later availability changes are listed under follow-up releases on this page.
Use mistral-small-2603 in Mistral AI's API.
Newsletter
Each month we pull Mistral AI's biggest releases into one clear read, alongside the launches that mattered from the rest of AI, so you don't have to watch every changelog yourself.
The standout launches, model drops, and platform moves worth tracking.
One tight read instead of bouncing between company blogs and timelines.
Clear summaries you can scan fast and forward to the team when needed.
Monthly dispatch
Get a clean monthly dispatch covering the launches that mattered, not every minor product breadcrumb.
One email each month, tuned for signal over noise.