The day in AI, in five minutes.
Mistral announced Mistral Large 4 at $0.68 in and $2.09 out per million tokens, while Google released EmbeddingGemma 2, an open-source multimodal embedding model. Claude entered the top 10 free apps on the US App Store at #9, Google DeepMind's embeddinggemma-2 ranked #10 trending on Hugging Face, and OpenAI's share of OpenRouter spend rose 1.2 points in a week to 28.4%. Service issues affected Anthropic's usage data and OpenAI's latest desktop app, with both status pages back to normal.
What mattered
- 1
Mistral releases Mistral Large 4
Why it matters: Mistral Large 4 gives builders a new model option with stated input and output prices of $0.68 and $2.09 per million tokens.
Mistral announced Mistral Large 4 on its official site.
- 2
Google releases EmbeddingGemma 2 open multimodal embedding model
Why it matters: EmbeddingGemma 2 lets builders map text, code, images, video and audio into a unified 768-dimensional space, with modular encoders from 270M to 740M parameters and truncation down to 128 dimensions.
Google released EmbeddingGemma 2, a compact open-source multimodal embedding model mapping text, code, images, video and audio into a unified 768-dimensional space. It offers modular encoders from 270M to 740M parameters and Matryoshka truncation down to 128 dimensions.
- 3
Mistral releases trillion-parameter open-weight model Le Chonk
Why it matters: Mistral reportedly released Le Chonk, a trillion-parameter open-weight model, and the company says it is the best open-weight offering outside of China.
Mistral released a new trillion-parameter model called Le Chonk. The company says it is the best open-weight offering outside of China.
- 4
Anthropic resolves usage data errors on Claude platform
Why it matters: Anthropic's resolved usage data errors affected the Usage page and Admin API usage report for about 73 minutes, relevant to people checking usage on the platform.
Anthropic reported elevated errors loading usage data on platform.claude.com. The incident, affecting the Usage page and Admin API usage report, lasted about 73 minutes and has been resolved.
- 5
OpenAI resets Codex and ChatGPT Work usage limits
Why it matters: OpenAI staff announced a reset of Codex and ChatGPT Work usage limits after community demands cited in the post.
OpenAI staff announced a reset of Codex and ChatGPT Work usage limits. The post says the community demanded a reset after four shipped items and some math proofs.
The Vikshy Score
The close of 7 Oct, 06:00 UTC, against the previous close. The full board.
The numbers
Read straight from the source. Every signal.