SandRise
Ep 965 Blog 0:52 w/ Edmund

AIM — India

The company has made V4.1-Flash available through its API with native multimodal support. DeepSeek has retired V4-Flash and V4-Flash-Vision-Exp, while the older model names deepseek-v4-flash and deepseek-v4-flash-vision-exp will temporarily route requests to V4.1-Flash.

Blog
Read the source →
Embed this episode

Paste this on any site — the player is a self-contained iframe with no cookies or trackers. 

<iframe src="https://sandrise.io/exploring-next/embed/965"
  width="100%" height="180" style="max-width:640px;border:0;border-radius:12px;overflow:hidden"
  title="Exploring Next — Episode 965 audio player"
  loading="lazy" allow="autoplay" referrerpolicy="strict-origin-when-cross-origin"></iframe>
Embed & API docs →
Script Built In Brief Voice Speechify Simba 3.2

Transcript

Edmund Quick look at AIM — India. Here is what is worth knowing.

Edmund the company has made V4.1-Flash available through its API with native multimodal support.

Edmund What is interesting here is its sparse attention is trained at a 64K sequence length, with the context window extended to 1 million tokens after training on 34 trillion tokens.

Edmund A card revision made after entry moved Param2-17B-A2.4B-Thinking's stated context ceiling from 4,096 tokens to 32,768, cutting the distance to the promised 128K by a factor of eight and making delivery by 31 October materially more plausible than it looked at entry.

Edmund If you want to go deeper, the source is worth reading — link's in the show notes.

Edmund That is the brief.