---
type: "post"
id: "introducing-mistral-large-4-le-chonk-analysis-ddedc3"
slug: "introducing-mistral-large-4-le-chonk-analysis-ddedc3"
title: "Introducing Mistral Large 4: Le chonk"
description: "Mistral Large 4’s preview couples very large stated scale with a staged access path: API availability now and promised open weights later."
retrieval_nugget: "Mistral Large 4’s preview couples very large stated scale with a staged access path: API availability now and promised open weights later."
published_at: "2026-10-06"
updated_at: "2026-10-06"
record_date: "2026-10-06"
date_kind: "published_at"
topics: []
source_urls: ["https://simonwillison.net/2026/Oct/6/le-chonk"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:introducing-mistral-large-4-le-chonk-analysis-ddedc3"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:d4d3157b647aadd77120a7a2d55f5a687dfe26779162374764ab4a0a5c174139","reviewed_at":"2026-10-06T21:00:57.209Z","source_evidence_count":1,"verified_claim_count":4,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":4,"implication_count":1,"watch_condition_count":1,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"The preview signals a staged return to frontier competition, but scale and one reported score do not establish parity.","observed_facts":[{"text":"The source says the API preview is available and describes the model as having 1 trillion parameters, 49 billion active parameters.","source_urls":["https://simonwillison.net/2026/Oct/6/le-chonk"]},{"text":"The source says training used Mistral’s own cluster of 3,800 NVIDIA Grace Blackwell GPUs.","source_urls":["https://simonwillison.net/2026/Oct/6/le-chonk"]},{"text":"The source reports two API reasoning settings, “none” and “high,” and an Artificial Analysis score of 38 versus 9 for Large 3.","source_urls":["https://simonwillison.net/2026/Oct/6/le-chonk"]},{"text":"The source says Mistral promises open weights at the end of the month; this is a future commitment, not a completed release.","source_urls":["https://simonwillison.net/2026/Oct/6/le-chonk"]}],"mechanism":"Mistral is offering API access before the promised open-weights release, while presenting a model with 1 trillion total parameters and 49 billion active parameters.","why_now":"A concrete preview is available through the API, while the company has set an “end of this month” expectation for open weights.","implications":["The split between API preview and later weights means access, reproducibility, and deployment options may change only if the promised release arrives with usable terms."],"evidence_boundary":"The supplied source is a commentator’s account of the release, and its single benchmark score cannot establish broad capability, cost, or frontier standing.","watch_conditions":["Check whether Mistral releases the weights on schedule, with what license, and how independent evaluations compare across tasks."],"related_records":[],"new_branch_reason":"No prior New Runtime record is supplied for this specific Large 4 preview; this starts a distinct record grounded in the reported API release and planned weights."}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/introducing-mistral-large-4-le-chonk-analysis-ddedc3/","markdown":"https://newruntime.com/posts/introducing-mistral-large-4-le-chonk-analysis-ddedc3.md","json":"https://newruntime.com/posts/introducing-mistral-large-4-le-chonk-analysis-ddedc3.json"}
---

# Introducing Mistral Large 4: Le chonk

## Retrieval answer

Mistral Large 4’s preview couples very large stated scale with a staged access path: API availability now and promised open weights later.

Mistral Large 4’s preview couples very large stated scale with a staged access path: API availability now and promised open weights later. That makes the release more than a parameter-count headline, while leaving the most consequential question—what users can actually run and under what terms—unresolved.

Thesis: The preview signals a staged return to frontier competition, but scale and one reported score do not establish parity. Mechanism: Mistral is offering API access before the promised open-weights release, while presenting a model with 1 trillion total parameters and 49 billion active parameters. Why now: A concrete preview is available through the API, while the company has set an “end of this month” expectation for open weights. The supplied account says training used Mistral’s own cluster of 3,800 NVIDIA Grace Blackwell GPUs. It also reports two API reasoning settings, “none” and “high,” and an Artificial Analysis score of 38, compared with 9 for Large 3.

Implication: The split between API preview and later weights means access, reproducibility, and deployment options may change only if the promised release arrives with usable terms. Evidence boundary: The supplied source is a commentator’s account of the release, and its single benchmark score cannot establish broad capability, cost, or frontier standing. Watch condition: Check whether Mistral releases the weights on schedule, with what license, and how independent evaluations compare across tasks.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)