Fetching from the wire…
01
02
03
04
05
06
07
08
Tools2026-09-17 · source-backed
cli-v0.3.0 cut September 17, and the repo gained 1,350 stars today to reach 3,604. The headline is a canvas visual-object-model track: discovering canvas candidates, resolving visible regions, typed visual refs with target identity validation, visible-region screenshots bound to a canvas ref, cursor continuation when there's too much to show, and point interactions on canvas regions by screenshot coordinate. Also real mouse-wheel, focus/blur and scroll-to-element primitives, host-managed daemons for sandboxed agents, and local task execution history for audit. Canvas has been the dead zone for DOM-driving agents, and coordinate interaction bound to a validated ref is a reasonable answer.
Each link below shares sources, entities, or timing with this story.
22,521 stars, 3,233 forks, 749 open issues, Go, gaining 226 stars on September 12. It bundles a queryable RAG store, an autonomous reasoning agent, and a wiki the model keeps updating as sources change (GitHub). Three repos on the same boards now take that shape: WeKnora, nash...
Released August 28 with 78 layers, 77 of them MoE with 256 routed plus one shared expert and top-8 routing, plus a native 10B MTP layer for speculative decoding (GitHub). The attention stack uses Gated DeepSeek Sparse Attention with IndexCache for cross-layer sparse index reus...
On July 14, llama.cpp merged native support for Tencent's Hunyuan Hy3 architecture (PR #25395), a 295B-parameter, 21B-active MoE. Any recent master build can load it now. Community GGUF quants (Q2_K, IQ2_M, Q4_K_M) from AngelSlim and others already ship on Hugging Face, and so...
Simon Willison ran it and notes the jump from Hy3's 295B/21B/256K, two reasoning effort levels with 'high' as default and 'no_think' available, and a reasoning trace written in slightly truncated English that reads as the model deliberating, considering and rejecting details l...
The AngelSlim/Hy4-preview-GGUF repo offers Q4_K_M at 435.20 GiB (4.86 bpw) and STQ1_0 at 213.66 GiB (2.38 bpw), benchmarked at 204.56 t/s prefill and 20.47 t/s decode on 8x H20. STQ1_0 comes from llama.cpp PR #22836 and uses ternary weights with 3:4 forced sparsity at 1.3125 b...
Hy4 preview, released August 28: 770B total parameters, 49B active, over 1M token context, open-sourced and simultaneously on Tencent Cloud TokenHub and OpenRouter at $0.834 per million input tokens, $2.501 per million output, $0.042 per million cached. In Tencent's own evalua...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.