Skip to main content
wideriver_

$ watch the frontier --daily --verified

Which model should you actually reach for? Verified scores from official releases and papers, news from seven primary sources — readings for people who build, tinker, and experiment on their own. No hot takes, no aggregator soup.

From the Desk

the Ledger

471
Models measured
2142
Articles tracked
claude-fable-5
Ledger leader
10
Tonight's wire
  1. 1
    claude-fable-5
    Anthropic · 1 benchmark
    100
  2. 2
    Stem Grad
    Human · 1 benchmark
    99
  3. 3
    Gemini 3 Deep Think (2/26)
    Google · 2 benchmarks
    99
  4. 4
    claude-opus-4-6-thinking
    Anthropic · 1 benchmark
    99
  5. 5
    GPT-5.4 Pro (xHigh)
    OpenAI · 2 benchmarks
    98

Method: mid-rank percentile within each benchmark's population · ties by verified count

Full ledger →

the wire

10 items on tonight's wire

Full wire →

The Evening Briefing, in your inbox by 7.

One quiet email a day — the verified movers and the wire that matters — read before your first coffee finishes.

We'll send you a confirmation email. No spam, unsubscribe anytime.