Evidence: benchmark results vendor-reported
Mistral Large 4 1T-A52B — VRAM, speed & local setup
Written by Jakub Rusinowski · Last updated
About 1T parameters in total and 52B active per token (49B without the embeddings and output layer): 128 routed experts with 4 active plus one shared expert, latent attention with a sparse-attention indexer, a 1M-token context and a 1.6B vision encoder. Reasoning effort is low, medium or high and cannot be turned off. Mistral says it trained the model from scratch on 3,800 NVIDIA Grace Blackwell GPUs in its own European data centres. Official weights are FP8 and NVFP4; the llama.cpp maintainers publish GGUFs from 298 GB (IQ2_XXS) to 2,099 GB (BF16).
Mistral Large 4 1T-A52B needs about 636 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
Get Mistral Large 4 1T-A52B running
Mistral Large 4 1T-A52B is not practical on a single consumer GPU: it needs about 635.6 GB at Q4_K_M, more than any single device in our catalogue holds with headroom. How multi-GPU inference works, or rent a multi-GPU server by the hour:
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
How to run Mistral Large 4 1T-A52B
Ollama lists Mistral Large 4 only as the cloud tag mistral-large-4:cloud, which runs on Ollama's servers, not your machine. Locally, use llama.cpp (mainline since 29 October 2026) with the ggml-org GGUFs, or vLLM / SGLang on a multi-GPU server.
Specifications
Verified — Checked against the primary source — the model card or the vendor spec page — and corroborated by a second independent source.
- Parameters
- ~1 Trillion (52B active)
- Context window
- 1,048,576
- Architecture
- Mixture-of-Experts (128 routed + 1 shared, 4 active), MLA with sparse-attention indexer
- Provider
- Mistral AI
- Licence
- Mistral Ecosystem Open-Weight License
- Availability
- Weights downloadable
- Practical deployment
- Private cloud / datacenter cluster
- Specified at
- Q4_K_M
- Vision projector
- +3.4 GB when you send images (not included above)
- System RAM
- 1024 GB
- Record updated
- 2026-11-03
Weights are downloadable and commercial use is permitted, subject to the licence’s acceptable-use terms.
Limits and caveats
- Datacentre-only in practice: every expert must be resident. The one single-machine route is a 2-bit GGUF (IQ2_XXS 297.7 GB or Q2_K 380.7 GB), which the llama.cpp maintainers say fits 512 GB of unified memory; on Apple Silicon that is close to the default GPU memory limit.
- No KV-cache or speed estimate here: the cache is shared across layers and a sparse-attention index adds its own, which our model does not yet express. Hugging Face measured about 15 tokens/s generation with Q2_K on one Mac Studio M3 Ultra (512 GB).
- On Mistral's API the model is still labelled Public Preview (docs version 26.10).
- Every benchmark in the launch material is vendor-reported; preview and final numbers differ and must not be mixed.
Quality and use cases
Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.
Mistral Large 4 1T-A52B — frequently asked questions
How much VRAM does Mistral Large 4 1T-A52B need?
About 636 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. We publish no context-length figure or speed for it: its attention shares the KV cache across layers and adds a sparse-attention index cache, which our memory and speed model cannot yet express, so we state the published weights only. Apple Silicon counts unified memory toward the same figure.
Does Mistral Large 4 1T-A52B run on an RTX 4090 (24 GB)?
Its weights and overhead come to about 636 GB at Q4_K_M, but we do not publish a fit verdict or a speed for it: its attention shares the KV cache across layers and adds a sparse-attention index cache, which our memory and speed model cannot yet express, so we state the published weights only.
How do I run Mistral Large 4 1T-A52B locally?
Ollama lists Mistral Large 4 only as the cloud tag mistral-large-4:cloud, which runs on Ollama's servers, not your machine. Locally, use llama.cpp (mainline since 29 October 2026) with the ggml-org GGUFs, or vLLM / SGLang on a multi-GPU server. Running the published tag would send your prompts to a hosted GPU rather than your own machine.