Mistral: upcoming dates

Recent

• AI10/12/2026@TechDesk

  Mistral Large 4 Released

  15 Rue des Halles, 75001 Paris, FranceMistral AI

  

  Starts 10/06/2026CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins84% CONFIDENCEOverall confidence: 84%How well the pin's source and references back up its dates.Weighted average of how firmly 5 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better11 days ago

  Mistral's^[1][5] announcement 'Introducing Mistral Large 4' is dated October 6, 2026 and describes a public preview; its docs model card lists the release date as October 6, 2026 (v26.10, Public Preview) and TechCrunch's^[2] report is dated the same day.

  Mistral AI releases Mistral Large 4 in public preview, a natively multimodal mixture-of-experts model of about 1 trillion parameters trained on its own cluster, with open weights promised by the end of October 2026.

  10
• AI10/12/2026@TechDesk

  Mistral Medium 3.5 Released

  15 Rue des Halles, 75001 Paris, FranceMistral AI

  

  Starts 04/29/2026CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins82% CONFIDENCEOverall confidence: 82%How well the pin's source and references back up its dates.Weighted average of how firmly 7 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better171 days ago

  The Hugging Face^[4] weights for Mistral^[1][7]-Medium-3.5-128B first appear on 29 April 2026 and creativeainews's^[5] 29 April report says 'On April 29, 2026, Mistral released Medium 3.5'; Mistral's docs changelog dates the API entry one day earlier (April 28) and its announcement page carries a later 22 May stamp, so 29 April is the public launch day.

  Mistral AI releases Mistral Medium 3.5, its first merged flagship: a 128B dense open-weight model with a 256K context window that scores 77.6% on SWE-Bench Verified and costs $1.50 per million input tokens and $7.50 per million output tokens.

  10
• AI10/12/2026@TechDesk

  Mistral Large 3 and Ministral 3 (3B, 8B and 14B) Released

  15 Rue des Halles, 75001 Paris, FranceMistral AI

  

  Starts 12/02/2025CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins81% CONFIDENCEOverall confidence: 81%How well the pin's source and references back up its dates.Weighted average of how firmly 6 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better319 days ago

  Mistral's^[1][6] announcement 'Introducing Mistral 3' is dated December 2, 2025 and its changelog says 'December 2: We released Mistral Large 3 (mistral-large-2512)'; TestingCatalog's^[5] report carries the same date.

  Mistral AI releases Mistral 3: Mistral Large 3, a 675B-parameter sparse mixture-of-experts model with 41B active parameters under Apache 2.0, and three dense Ministral 3 models (3B, 8B and 14B).

  30
• AI10/12/2026@TechDesk

  Mistral Small 3.2 Released

  15 Rue des Halles, 75001 Paris, FranceMistral AI

  

  Starts 06/20/2025CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins82% CONFIDENCEOverall confidence: 82%How well the pin's source and references back up its dates.Weighted average of how firmly 5 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better1.3 years ago

  Mistral's^[1][3] model card for mistral-small-2506 lists the release date as June 20, 2025 and its changelog says 'June 20: We released Mistral Small 3.2' with the API following on June 23; VentureBeat's^[5] report is dated 20 June 2025.

  Mistral AI releases Mistral Small 3.2, a 24B Apache 2.0 update to Small 3.1 that lifts Arena Hard v2 from 19.56% to 43.1% and cuts infinite generations from 2.11% to 1.29%.

  20
• AI10/12/2026@TechDesk

  Mistral Small 3.1 (24B) Released

  15 Rue des Halles, 75001 Paris, FranceMistral AI

  

  Starts 03/17/2025CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins88% CONFIDENCEOverall confidence: 88%How well the pin's source and references back up its dates.Weighted average of how firmly 5 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better1.6 years ago

  Mistral's^[1][5] announcement 'Mistral Small 3.1' is dated March 17, 2025 and its changelog says 'March 17: We released Mistral Small 3.1 (mistral-small-2503)'; Maginative's^[4] report carries the same date.

  Mistral AI releases Mistral Small 3.1, a 24B Apache 2.0 model with image understanding and a 128k-token context window that Mistral says beats Gemma 3 and GPT-4o Mini at 150 tokens per second.

  20
• AI10/12/2026@TechDesk

  Mistral Small 3 (24B) Released

  15 Rue des Halles, 75001 Paris, FranceMistral AI

  

  Starts 01/30/2025CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins88% CONFIDENCEOverall confidence: 88%How well the pin's source and references back up its dates.Weighted average of how firmly 5 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better1.7 years ago

  Mistral's^[1][5] announcement 'Mistral Small 3' is dated January 30, 2025 and its changelog says 'January 30: We released Mistral Small 3 (mistral-small-2501)'; Simon Willison's^[3] post carries the same date.

  Mistral AI releases Mistral Small 3, a latency-optimized 24B model under Apache 2.0 that it says matches Llama 3.3 70B instruct at more than three times the speed, with over 81% on MMLU.

  10
• AI10/12/2026@TechDesk

  Ministral 8B and Ministral 3B Released

  15 Rue des Halles, 75001 Paris, FranceMistral AI

  

  Starts 10/16/2024CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins80% CONFIDENCEOverall confidence: 80%How well the pin's source and references back up its dates.Weighted average of how firmly 5 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better2.0 years ago

  Mistral's^[1][5] announcement 'Un Ministral, des Ministraux' is dated October 16, 2024, the Ministral-8B-Instruct-2410 repo was released as 'Release 241016' that day, and The Decoder^[3] reported it on 16 October; Mistral's docs changelog lists the API models one week earlier, under October 9.

  Mistral AI releases Ministral 8B and Ministral 3B, 'les Ministraux', edge models with up to 128k context; the 8B scores 65.0 on MMLU (pretrained) and 70.9 on Arena Hard, priced at $0.10 and $0.04 per million tokens.

  50
• AI10/12/2026@TechDesk

  Mistral Large 2 Released

  15 Rue des Halles, 75001 Paris, FranceMistral AI

  

  Starts 07/24/2024CONFIRMEDDate confidence: CONFIRMEDStated as firmHow firmly the source words the date: confirmed, scheduled, estimated, delayed or unverified.Show all CONFIRMED pins80% CONFIDENCEOverall confidence: 80%How well the pin's source and references back up its dates.Weighted average of how firmly 6 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better2.2 years ago

  Mistral's^[1][6] announcement 'Large Enough' is dated July 24, 2024 and its changelog says 'July 24: We released Mistral Large 2 (mistral-large-2407)'; TechCrunch^[4] published the same day.

  Mistral AI releases Mistral Large 2, a 123B model with a 128k-token context that it says matches GPT-4o and Claude 3 Opus on code and beats Llama 3.1 405B on code and math with under a third of the parameters.

  20