From the source

Introducing Mistral Large 4: Le chonk

Mistral releases preview of 1T-parameter model with open weights coming soon

New model

Mistral released a preview of Mistral Large 4, a 1 trillion parameter model with 49 billion active parameters, trained on its own cluster of 3,800 NVIDIA Grace Blackwell GPUs.

The preview is available via the Mistral API and supports two reasoning levels: 'none' and 'high'.

Open weights are promised by the end of the month.

From the source
Today they're releasing a preview of Mistral Large 4, a 1 trillion parameter, 49 billion active parameter model trained on their own cluster of 3,800 NVIDIA Grace Blackwell GPUs.
simonwillison.net

Other stories

Hugging Face · 7 OctNew model

Introducing Falcon ASR

Falcon-ASR is a 1.6B parameter Arabic speech recognition model with Emirati dialect focus.

Anthropic · 7 OctNew model

Introducing Claude Haiku 5.5

Anthropic releases cheaper, faster Haiku 5.5

Mistral AI · 6 OctNew model

Announced: 06 Oct 2026 20:18 UTC
Detected: 06 Oct 2026 20:30 UTC (+12m after)

ModelsMistral Large 4

Visit source ↗simonwillison.net

Introducing Mistral Large 4

Mistral Large 4 previews with 1T-parameter open-weight model, weights due this month.

Hugging Face · 6 OctNew model

Falcon-Emirati: When an LLM Learns the Dialect, the Culture, and the Nuance

Falcon-Emirati-7B adapts a general Arabic model to the Emirati dialect and culture.

Multimodal embeddings beyond a single vector
New late-interaction embedding models with multi-vector representations and multimodal support.
Perplexity · 7 OctNew model
Introducing Claude Haiku 5.5 on AWS
Anthropic's fastest Claude model arrives on AWS with cost savings and effort controls.
Amazon · 7 OctAvailability change

Introducing Mistral Large 4: Le chonk

From the source

Introducing Falcon ASR

Introducing Claude Haiku 5.5

Falcon-Emirati: When an LLM Learns the Dialect, the Culture, and the Nuance

Multimodal embeddings beyond a single vector

Introducing Claude Haiku 5.5 on AWS