1.0The product
Now accepting new users

Your own research desk,not another aggregator.

Curate the newsletters you trust. We consolidate them into one clean, deduplicated brief — delivered where you want, on your schedule.

synthwire.app / digest / today

Your brief · Today

11 stories · 6 min · saved 47 min

June 13, 2026

From 9 sources you trust

Signal
📈⚖️⚠️3 sent.
AI-curated · 0.94

🤖Frontier Models & Research

2 stories

Three significant releases this week: DeepMind's AlphaGeometry2 achieves IMO gold-medal performance, new open-weight model family challenges GPT-4o on coding benchmarks, and scaling law research suggests compute efficiency gains are accelerating.

I

from Import AI

3 min · 96% match

AlphaGeometry2: IMO Gold, Open-Weight Challengers, Scaling Law Updates

DeepMind's AlphaGeometry2 solves 42/50 IMO geometry problems (gold medal threshold). Separately, Z.ai releases GLM-4-9B-0414 beating GPT-4o on HumanEval. Epoch AI finds compute efficiency doubling every 8 months — faster than Moore's Law.

🏭Semiconductors & Compute

2 stories

TSMC 2nm risk production on track for H2 2025; US CHIPS Act funding flowing to AZ fabs; new memory architecture could reduce AI training energy by 40%.

S

from Semiconductor Daily

2 min · 94% match

TSMC 2nm Yield Improving, Arizona Fab 21 Breaks Ground

TSMC reports 2nm yield at 60%+ in risk production. Second Arizona fab begins construction with $6.6B CHIPS Act funding. Samsung 2nm GAA facing yield challenges.

⚖️Crypto Policy & Regulation

1 stories

Stablecoin bill advances in Senate; EU MiCA implementation enters Phase 2; SEC drops several enforcement actions post-election.

C

from Crypto Policy Review

3 min · 92% match

GENIUS Act Markup: Stablecoin Framework Takes Shape

Senate Banking Committee advances stablecoin bill with federal/state dual charter system. Reserve requirements, audit standards, and custodial rules defined. Industry response mixed.

11 stories

6 min read

saved 47 min

Sent to email · Slack

Synthwire · v1.0

1.1Indexing 10,000+ vetted newsletters

The sources you already trust.

I
Import AIweekly
S
Stratecheryweekly
S
Semiconductor Dailydaily
C
Crypto Policy Reviewweekly
G
Game Developer Digestweekly
T
The Batchweekly
C
ChinAI Newsletterbiweekly
A
AI Safety Newslettermonthly
T
TLDR AIdaily
L
Latent Spaceweekly
T
The Economist - Technology Quarterlyquarterly
C
Construction Physicsweekly
I
Import AIweekly
S
Stratecheryweekly
S
Semiconductor Dailydaily
C
Crypto Policy Reviewweekly
G
Game Developer Digestweekly
T
The Batchweekly
C
ChinAI Newsletterbiweekly
A
AI Safety Newslettermonthly
T
TLDR AIdaily
L
Latent Spaceweekly
T
The Economist - Technology Quarterlyquarterly
C
Construction Physicsweekly
1.2How it works

Three steps. No magic.

Other tools promise AI auto-discovery and fail at the inbox-confirmation step. Synthwire is honest: you do the subscribing, we do the reading.

01

Curate

You pick the sources.

Browse a directory of 10,000+ vetted newsletters. Filter by topic, frequency, paywall, or author type. One click to add.

02

Subscribe

You confirm, like always.

Clicking "Subscribe" opens the newsletter's normal signup. You confirm via the standard double-opt-in. Your email stays yours.

03

Receive

We consolidate, you read.

Ten newsletters covering the same story become one. Deduplicated, summarized, ranked. Delivered daily to email, Slack, Discord, or SMS.

1.3How a digest gets built

Seven newsletters cover the samestory.

One synthesis. One read. Six minutes instead of forty. Same coverage. None of the inbox archaeology.

  • S
    stratechery.comcovers this story

    OpenAI’s new model changes the unit economics of inference

    The cost-per-token has fallen below a threshold that makes always-on agents viable for the first time. Implications for SaaS pricing are significant.

    8 min

  • I
    import-ai.newscovers this story

    Jack Morris — what an order of magnitude cheaper inference means

    Cheaper inference reframes the question from "can we afford to run this model?" to "what should we run it on?" The bottleneck moves to evaluation.

    6 min

  • T
    theinformation.comcovers this story

    Inside OpenAI’s aggressive GPU reserve strategy for 2026

    OpenAI is locking in multi-year compute commitments, signalling it expects inference demand to outpace training within 18 months.

    7 min

  • T
    techcrunch.comcovers this story

    OpenAI announces 90% price cut on its flagship model

    The new pricing undercuts every major competitor and accelerates a race to the bottom for inference APIs.

    4 min

  • T
    theverge.comcovers this story

    OpenAI just made its smartest model basically free

    For consumers, the price drop is invisible. For developers building products, it changes the math on what is possible to ship this quarter.

    5 min

  • W
    wired.comcovers this story

    The quiet collapse of the AI compute moat

    When the marginal cost of intelligence falls by an order of magnitude, the value shifts from owning models to owning distribution and data.

    6 min

  • P
    platformer.newscovers this story

    What OpenAI’s price cut tells us about the agent economy

    The strategic move is not about API revenue. It is about making agents ubiquitous so the next platform layer belongs to OpenAI.

    5 min

Synthwire synthesis

live

OpenAI’s 90% inference price cut signals a shift from training to agent economics.

On Monday, OpenAI cut API prices by 90% — an aggressive move that makes always-on agents economically viable for the first time. Seven trusted sources covered the same story from different angles: pricing, strategy, GPU reserves, and the downstream collapse of the compute moat. The synthesis below captures the consensus and the disagreement.

Covers

stratechery.com · import-ai.news · theinformation.com · techcrunch.com · theverge.com · wired.com · platformer.news

Signal

quality0.94

Real product output from a beta user · OpenAI price cut, June 2026

demo · static · no state

1.4Why us

The other newsletter
tools, then us.

We don't pretend to read your inbox, bypass your newsletter confirmations, or invent sources you didn't ask for. We do the part AI is actually good at: consolidation.

Property
Other tools
Synthwire
Who chooses sources
Black-box AI
You do
Inbox crawling
Often required
Never
Captcha handling
Pretends to (doesn't)
Not needed
Trust model
Opaque algorithm
Your source list
Email exposure
Sometimes
Never
Deduplication
Basic
Semantic + source attribution
Reading modes
One
Executive / Analyst / Deep
Cancellation
Hunt for links
One click
1.6Proof, not promises

The numbers that actually matter.

01

87%

Duplicate Reduction

Stories collapsed across sources

02

4.2 min

Median Read Time

Per daily digest

03

73%

Sources Kept at 30 Days

Retention vs 31% industry avg

04

3.4x

New Ideas per Digest

vs manual curation

1.7Features

You curate the sources. We do the work.

Most newsletter tools try to auto-discover for you and fail at trust. Synthwire flips the model: you own the source list, we handle the reading.

You choose every source

No AI auto-subscribing. Browse a directory of 10,000+ vetted newsletters and add the ones you want. We never subscribe on your behalf.

Source similarity

"People who read these also kept these 3 sources" — find writers like the ones you already trust.

Redundancy filter

Ten newsletters covering the same story → one entry. 87% duplicate reduction.

Coverage gaps

"You follow AI policy but miss 3 high-signal semiconductor sources." Blind-spot detection on your curated list.

Full source transparency

Every source shows publisher, topics, cadence, paywall status, reputation, and overlap with your other sources.

Topic memory

"Show me only new ideas, not repeated takes." Filter rehashed narratives across your sources.

Reading modes

Executive (3 min), analyst (8 min), deep-research (20 min). Match depth to available time.

You control routing

Email, Slack, Discord, or SMS. Pause all sources. One-click unsubscribe. Export anytime.

Privacy first

No captchas. No inbox crawling. No third-party authentication. You own your data — export or delete with one click.

1.8Pricing

Pay for what you read.

No team seats you don't need. No enterprise tier that hides the good stuff. Pick the level that matches how you work.

01

Free

Try the experience.

For the curious.

$0/mo
  • 5 sources
  • 7-day archive
  • Email delivery
Start free
Popular

02

Lite

Build a real feed.

For the practitioner.

$3/mo
  • Unlimited sources
  • 90-day archive
  • Email delivery
  • Slack delivery
  • Redundancy filter
  • Coverage gap detection
  • Source graph
Get Lite

03

Max

The full research desk.

For teams and analysts.

$7/mo
  • Unlimited sources
  • 365-day archive
  • Email delivery
  • Slack delivery
  • Redundancy filter
  • Coverage gap detection
  • Source graph
  • Full-text archive search
  • On-demand source requests
  • Team workspaces (10 members)
Get Max

No credit card required · Cancel anytime

1.9FAQ

Questions, answered.

Can't find what you're looking for? Email us.

Right — the AI can't bypass confirmation emails or captchas. So we don't try. You browse our curated directory of 10,000+ vetted newsletters, click "Subscribe" on the ones you want, and confirm via their normal signup flow. Synthwire's role is consolidation: we read what you subscribed to, deduplicate, summarize, and route a clean daily brief.

Three options. (1) Most common: subscribe at your normal email, and the ones you pick also get a copy forwarded to your unique @synthwire.email address. (2) Direct: sign up with your Synthwire address as primary. (3) Manual forward: forward any newsletter to your Synthwire address. Your real email is never exposed (option 1) and you can revoke any source instantly.

Everything. Pause any source instantly. Mute for 30 days. Remove permanently. Adjust reading mode per-source. Filter by topic, source type, or author. Export your full data as JSON anytime. Delete your account with one click — we purge all data within 24 hours.

When multiple sources cover the same story, we cluster them by semantic similarity. You see one synthesized entry with the best summary, attributed to all covering sources. Original links preserved. ~87% duplicate reduction based on beta data.

Executive brief: 3-sentence summaries, ~3 min. Analyst brief: structured sections with context, ~8 min. Deep-research: full context with citations, ~20 min. You can set per-topic — deep for AI models, executive for crypto policy.

We track the semantic fingerprint of every idea across your feed. When a new issue arrives, we compare its key claims against your history. If it's a rehash of a narrative you've already seen, it's flagged or filtered. You see "new idea" vs "repeated take" labels. Tunable sensitivity.

Yes (Max tier). Full-text search across every indexed issue with filters: source, date range, topic tags. Results ranked by relevance with highlighted excerpts.

Max tier includes shared queues: a curated feed of sources your team agrees on, with routing policies (all items, curated-only, voted). Members can propose sources, comment on items, and get a shared digest. Up to 10 members.

Free tier gives you 5 sources and 7-day archive — enough to test the experience. No credit card. Upgrade anytime. Yearly billing saves ~17%. Cancel anytime.

2.0Get started

Stop reading 47 newsletters.Read one brief.

Free while we're building. No credit card, no commitment, no AI crawling your inbox.

No captchas·No inbox crawling·No third-party auth·You own your data
Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·Synthwire·