Issue 006 — window: 24h to 06:00 UTC, 11 Aug 2026nofeed.dev
No. 006 · 11 August 2026 · 4 min← No. 005 · No. 007
SIGNALELEVATED|SHIPPED 5|SKIPPED 6

The one thing

01

Meta ships Muse Glimmer: 30B, Apache 2.0, built to run local agents on one GPU.

Muse Glimmer is a dense ~29.6B multimodal model from Meta Superintelligence Labs, open-weighted under Apache 2.0 and on Hugging Face today.1 It is trained around the agent loop — plan, call tools, check results, recover from failure — not chat. Quantized weights target under 20 GB so the model, KV cache, perception encoder and a DFlash speculative drafter fit a 24–32 GB envelope; Meta claims agentic reliability holds at 4-bit on that class of machine.12

Day-zero runtimes are already moving: transformers v5.15.0 adds the model; Ollama 0.32.7/0.32.8 adds MLX then NVIDIA/AMD paths; Hugging Face documents llama.cpp, vLLM and SGLang support landing with the weights.345 Integrations for Ollama, LM Studio and Unsloth are rolling this week rather than all live at publish time.1

VERDICT · WAIT — weights and HF load path are real; treat consumer-GPU speed claims as vendor until you time a tool loop on your box. Pin a digest, run one agent scaffold you already trust, measure tokens/sec and tool-call success before you move anything off the cloud default.
Deeper — what is open, what is not
  • Apache 2.0 covers full-precision BF16, 4-bit quants, the DFlash drafter and the ~1.8B perception encoder. Training data and training code are not released.2
  • Context is stated at 131k+ tokens; knowledge cutoff 4 Jan 2026; text+image in, text out; 100+ languages; controllable effort via system prompt.2
  • This is Meta's first fully open weights release since Muse Spark went proprietary. The licence is the news as much as the size.6

Shipped

05
Use it

Claude Sonnet 5 · $2/$10 made permanent

Anthropic's Sonnet 5 launch post was edited 10 Aug: introductory API pricing of $2 per million input tokens and $10 per million output stays. The $3/$15 step that was due 1 Sept is cancelled.7 Budget the locked rate. The platform pricing docs still show the old two-row table with an August 31 cutoff — treat the dated edit on the news page as the primary until docs catch up.8

MATTERS TO · anyone whose agents default to Sonnet 5 and who had a 50% step-up in the September forecast
nofeed.dev/issues/2026-08-11/sonnet-5-price-permanent/
Breaking

OpenAI Agents Python 0.20.0

Implicit default model is now gpt-5.6-luna. Explicit model strings, run-level overrides and OPENAI_DEFAULT_MODEL still win. Local MCP supports Python SDK v1 and v2; apps with custom MCP HTTP auth or client factories must use the HTTP types of the installed MCP major, or pin mcp<2.9

MATTERS TO · anyone who depended on the previous default model without setting one
nofeed.dev/issues/2026-08-11/openai-agents-python-0-20/
Use it

vLLM 0.27.0

Kimi K3 full stack in one release, Qwen3.5 dense/MoE, PyTorch 2.13 (breaking env), deeper FlashAttention 4 on SM100, and a pile of DeepSeek-V4 serving wins. 561 commits, 242 contributors.10

MATTERS TO · self-hosters pinning a serving engine this week, especially Kimi K3 or DeepSeek-V4 traffic
nofeed.dev/issues/2026-08-11/vllm-0-27/
Use it

Vercel Sandbox · Managed Images

Sandbox SDK v3 defaults to vercel/sandbox/universal:latest on Ubuntu 26.04 with Node 24, Python 3.14, and preinstalled opencode, claude-code, codex and pi. Old runtime stays as a deprecated alias; Amazon Linux stays only via runtime. Source images live in the public vercel/sandbox repo.11

MATTERS TO · anyone spinning agent sandboxes on Vercel who was baking tools at boot
nofeed.dev/issues/2026-08-11/vercel-sandbox-managed-images/
Use it

Cline v4.1.8

Yolo Mode toggle removed — it was cosmetic; auto-approve is now the single control, and prior Yolo setups migrate to approve-all. Vertex accepts any model ID by hand, including Fable 5, and stops applying Anthropic list prices to Vertex Fable bills.12

MATTERS TO · Cline users who thought Yolo did something, and anyone on Vertex
nofeed.dev/issues/2026-08-11/cline-4-1-8/

Promised, not shipped
Muse Glimmer partner runtimes (Ollama beyond first paths, LM Studio, Unsloth) and optimized llama.cpp / MLX / ExecuTorch — rolling this week · Muse Spark 1.2 open weights — Zuckerberg said "soon", no date · Claude Code auto mode default — 14 Aug 2026 (issue 005)

The conversation

01
The claim — including ours

A 30B Apache model aimed at always-on local agents reopens the question the cloud defaults closed: what workload still needs a metered frontier call once tool-use quality is good enough on one GPU.1

From the floor
  • Meta research blog · primaryVendor claim, dated

    Positions Glimmer against Gemma4-31B and Qwen3.6-27B on agentic benches (DeepSearch QA, MCP-Atlas, τ-Bench, SWE-Bench), and ships DFlash block drafting so generation stays interactive on M4/M5 Max and RTX 5090 class boxes.1

  • @aussieguy1234 · Hacker NewsReader, unverified

    Notes SWE-bench-verified in the Opus-of-not-long-ago band and asks the useful question: cloud is faster and more reliable for most software — do you need that margin for this task?13

  • @hougaard · Hacker NewsField report, unverified

    Ran the full build (~120 GB RAM), fed defective code, watched it loop and dig in. One session is not a bench; it is a reminder that agentic claims die on recovery, not on the first tool call.13

Our take

Open weights under Apache matter more than the parameter count: you can fine-tune, ship, and keep the stack when a lab changes its mind. That is the difference from Muse Spark.

The decision is not "local vs cloud". It is which loops are private or cost-sensitive enough to pin a 24 GB box, and which still buy frontier recovery. Measure your own scaffold before you rewrite the default.

GitHub releases: 28 watchlist repos polled, sweep complete (public API). Vendor feeds: 12 sources, sweep complete, 16 in-window. HN above the noise floor with thread comments. Fresh Reddit pulse (6 subreddits, 8 posts above floor). X accounts digest last collected 05:30 UTC 10 Aug (~24h old) and X keyword digest same age — background only, stated. No 11 Aug X digest in-repo.

Skip this

06

Everything we saw

78
78 candidates scanned · 14 used in this issue — the rest, with the reason each one was left out
The receipt. Ranking is only trustworthy if the discarded pile is visible, so here it is: everything the collectors surfaced in the window, with its signal and what we did with it.
ItemSourceSignalCall
Muse Glimmer open weightsresearch.meta.ai1083p · 592c HNled the issue
Muse Glimmer model cardhuggingface.coprimary cardone thing
Sonnet 5 pricing made permanentanthropic.comedit 10 Augshipped
OpenAI Agents Python v0.20.0github releases2.9hshipped
vLLM v0.27.0github releases8.8hshipped
Vercel Sandbox Managed Imagesvercel changelog12hshipped
Cline v4.1.8github releases2.1hshipped
transformers v5.15.0github releases19.6hsupport path for Glimmer
Ollama v0.32.8github releases6.3hGlimmer runtime
Vercel: network boundary essayvercel.com6hcontext for sandbox, not a separate ship
Claude Code v2.1.227github releases7.1hflag/Bash fixes
LiteLLM v1.96.0github releases7.7hproxy fixes + cosign note
Zuckerberg FT on open modelsHN 453ppaywalled colourbackground to Glimmer
Reddit pulse — Sonnet pricing, Cursor token waste, Muse Spark 1.2 rumourreddit pulse8 posts, 6 subspricing confirmed via Anthropic primary
X digests 10 Aug 05:30digests~24h oldbackground only
EOF

End of feed. That is everything from the window worth your time.
Next issue tomorrow, 06:00 UTC — and if nothing ships, it will say so in two hundred words.

Still developingClaude Code auto mode default lands 14 Aug (issue 005). Agent Plugins 1.0.0 (issue 004) — still waiting on a second shipping client. Anthropic pricing docs still list the cancelled Sonnet 5 Sept step — re-check when docs match the news edit.
Sources — each with the date it was read01 research.meta.ai — Muse Glimmer (2026-08-11) · 02 huggingface — Muse-Glimmer-30B card (2026-08-11) · 03 github — transformers v5.15.0 (2026-08-11) · 04 github — ollama v0.32.8 (2026-08-11) · 05 huggingface blog — Muse Glimmer day-0 (2026-08-11) · 06 venturebeat — Apache 2.0 context (2026-08-11) · 07 anthropic.com — Sonnet 5, Aug 10 pricing edit (2026-08-11) · 08 docs.anthropic.com — pricing (docs lag) (2026-08-11) · 09 github — openai-agents-python v0.20.0 (2026-08-11) · 10 github — vllm v0.27.0 (2026-08-11) · 11 vercel changelog — Managed Images (2026-08-11) · 12 github — cline v4.1.8 (2026-08-11) · 13 news.ycombinator.com — Muse Glimmer thread (2026-08-11)

Get it by email.

One issue every weekday. The whole thing, not a teaser.

Or RSS, if you would rather we never had your address.