Tasks and inputs
Mistral Large 4 is listed for Image-text-to-text, Text generation. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
SyncDev model directory · Profile updated Oct 13, 2026
Mistral Large 4 is an open-weight multimodal model released in public preview on the publisher’s announced release, for general reasoning, coding and agent applications.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: The current model page lists promotional USD rates of $0.68/M input, $0.07/M cached input and $2.09/M output, alongside regular rates of $1.36, $0.14 and $4.18. The model is in public preview; confirm the current promotion and deployment terms.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
Mistral Large 4 is an open-weight multimodal model released in public preview on the publisher’s announced release, for general reasoning, coding and agent applications.
Mistral documents 1.05 trillion total parameters and 52 billion active parameters, plus a vision encoder. The active count describes the experts used for a token; the total count still matters when loading and storing the weights. A 52B active count does not make this a 52B download.
The documented context allowance is one million tokens. Function calling, structured responses and document question answering make it a candidate for an assistant that works across code and supplied documents. Start with a failing repository issue or a document question whose answer you can verify. Include contradictory passages and tasks that require declining to invent missing facts.
For local deployment, confirm the exact weight licence and supported serving engine before sizing hardware. For a hosted route, record the API model ID, processing tier and rate limits. These choices change the operational comparison even when the underlying model family is the same.
The publisher currently displays promotional rates of $0.68 input and $2.09 output per million tokens. Treat the promotion as a dated offer. Measure the full cost of an accepted result, including repeated attempts, rather than estimating a budget from the first response alone.
Publisher source 1 Deployment limits and commercial terms belong to the selected checkpoint or API route.
The current model page identifies the 6 October release as a public preview, rather than a fully released production revision. It lists 1.05T total parameters, with 52B active and a 1.6B vision encoder. Promotional and regular API prices differ; check which rate applies on the deployment date. Open-weight positioning does not establish that a specific checkpoint download is available today.
Compare the model on the work you need to complete.
Mistral Large 4 is listed for Image-text-to-text, Text generation. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Documented limits: 1M context. The listed route has starting rates of $0.68 input and $2.09 output per million tokens. Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
The weights are downloadable under Publisher terms. Check runtime support and size the complete model and working context for your expected concurrency.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
Anthropic
Claude’s small model for high-volume extraction, classification and routing, with adaptive thinking and a 1M-token context window.
OpenAI
GPT-6.1 Sol is OpenAI's updated Sol model for coding and professional tasks. It accepts text and images and returns text.
inclusionAI
Ling 3.1 Flash is inclusionAI’s hybrid reasoning model for coding and tool agents, available through a hosted AI Gateway route.
Anthropic
Claude Sonnet 5.5 is an Anthropic language model that accepts text and images and returns text.
MiniMax
MiniMax M3.1 Flash Preview is a multimodal reasoning model for code and agent workflows, accessed through eligible MiniMax plans and tools.
Meituan
LongCat-2.5-Preview from Meituan: Meituan's multimodal reasoning model for coding and agent tasks, with image understanding and a 1M-token context window.