Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-08-23 · high
The Contributor tier undercuts Meta's standard pricing by up to 21 times, but the fine print lets Meta reuse everything sent through it.
Why now: The Contributor tier's price only became comparable across regions once OpenRouter listed it globally, per the site, giving builders a new number to weigh against the data terms.
Meta priced its Muse Spark 1.2 Contributor tier at $0.10 per million input tokens globally on OpenRouter, per the listing. The model card says prompts and outputs sent through it may be used to improve Meta's products, a straightforward trade of price for data. That disqualifies the tier for anything moving under a customer contract or a privacy promise.
The Contributor rate runs about 12.5 times cheaper on input and 21 times cheaper on output than Meta's standard Muse Spark 1.2 tier. Cached reads cost $0.002 per million tokens.
OpenRouter's listing shows 99.98% uptime over three days. Meta's terms separately tell users not to submit sensitive, confidential, or personal data through the tier.
That's fine for public-data work, side projects, or anything you wouldn't mind folded into a future Meta model. It doesn't work for anything under a customer contract or a privacy policy promising data stays put.
Anyone considering it for production work needs to know exactly what Meta counts as sensitive or confidential. The listing doesn't spell that out beyond the two words.
Each link below shares sources, entities, or timing with this story.
$1.25 input / $4.25 output per million tokens for the standard tier. $0.10 / $0.20 for the Contributor tier, where Meta trains on your usage and feedback. That's roughly 12x on input, 21x on output, and it's the clearest number anyone has published on what your proprietary sou...
For about a year, "run your agent locally" meant accepting a model that couldn't reliably call a tool twice in a row. That excuse is gone. Meta Superintelligence Labs published Muse Glimmer today: a 29.6B dense causal transformer, 52 layers, 6,656 hidden dim, with a ~1.8B ViT-...
The essay argues concentrating superintelligence in a few labs is the real risk, organized around individual empowerment, invention, and balance of power. Concrete commitments: a "personal superintelligence" agent for every user, an America's Workforce Academy with free trades...
Meta formally pivoted from open-weight Llama to fully proprietary Muse Spark, its first model from the newly formed Meta Superintelligence Labs. No downloadable weights. No self-hosting. Cloud-only private API preview to select partners. More locked down than OpenAI or Anthrop...
The NYT reported on July 17 that the June 2026 proposal is structured as monthly installments with an early-exit clause for either side, and would sit alongside Anthropic's existing $45B three-year SpaceX GPU deal from May. Meta fell about 6% intraday before closing down 2%. T...
Go look at your ~/.claude/CLAUDE.md right now. Mine has internal package names, a build command with a host in it, and notes about which credentials live where. I wrote it assuming exactly one reader. RuntimeWire published traced request captures on August 9 showing Muse Code...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.