Show HN: Cactus Needle 3: 8-29MB automation models can match DeepSeek V4 Flash

DeepSeek-v4.1 Flash: Pushing the Limits of KV Cache Compression

DeepSeek v4.1 flash runs 23 seconds/token on a 2020 16gb M1 Mac Mini

Turning GLM-5.3-Flash into a Jev-like decision model

Google Releases Gemini 3.8 Flash, Its Third Flash Model In Six Weeks

(Tech Report) DeepSeek-v4.1-Flash: Pushing Limits of KV Cache Compression [pdf]

I used AI pair-programming to write a clean-slate bare-metal Rust OS for a $40 RC transmitter (STM32F072, no_std, 128KB Flash)

Show HN: Running 104GB Qwen3.8-Flash-Next on 48GB Mac with at ~12 tok/s

BYD Luxury EV's Battery Capacity Only Dropped 1.3% After Flash Charging 350 Times

Gemini 3.7 Flash

Gemini Omni 1.1 Flash

The mechanics of the Nepali flash flood

Hot Chips 2026: Applying High Bandwidth Flash (HBF)

Gemini 3.7 Flash

DeepSeek V4 Flash 0731

Sandisk and SK Hynix Publish First Open High Bandwidth Flash Standard

DeepSeek V4 Flash 0731: 82.7% on Terminal-Bench 2.1 with a public harness

A new hybrid weather-and-evaporation model flags flash drought onset, persistence and end up to 10 days ahead, catching over 70% of cases across the US and Europe.

What GLM-5.3 Flash running on Chinese hardware actually means

Kioxia and Sandisk demonstrate the world's highest-density 3D NAND flash — 332 active layers and up to 4,800 MT/s interface

Qwen3.8-Flash-Next Intelligence, Performance and Price Analysis

DeepSeek V4 Flash on a Single AMD MI300X

GLM-5.3-Flash

Measured flash, RAM and time budgets for post-quantum signature verification in no_std Rust

GLM-5.3-Flash-GGUF

Simon Willison on DeepSeek-V4-Flash-0731

DeepSeek-V4-Flash-Vision-Exp

Qwen3.8-Flash-Next Technical Report [pdf]

Qwen3.8-Flash-Next: A New Architecture, Towards Ultimate Cost-Efficiency

Qwen3.8-Flash-Next

More →