Live telemetry · daily brief

GAWK

Nativerseby nativerse

ARCHIVE

2026-07-15

3 tool incidents

What verifiably moved in the AI ecosystem in the last 24h. Every number traces to a public source.

What moved

  • claude-opus-4-6-thinking holds #1 on LMArena for the 30th consecutive snapshot.
  • diffusers downloads grew for the 4th consecutive snapshot.

SDK adoption6 notable shifts across the six registries

Package install volume is where developers place real bets — distinct from where labs are shipping marketing. pypistats.org, npmjs.com, crates.io · view all →

−373.1k

PyPI: langchain

−373.1k 24h downloads day-over-day

pypistats.org is a third-party aggregator — counts include mirror hits, CI builds, and pip install retries.

PyPI · View →

+2.1M

npm: openai

+2.1M 24h downloads day-over-day

npm · View →

+36.5k

crates.io: ort

+36.5k 90d downloads day-over-day

crates.io · View →

+336.5k

Docker Hub: ollama/ollama

+336.5k all-time pulls day-over-day

Docker Hub · View →

+501

Homebrew: ollama

+501 30d downloads day-over-day

Homebrew · View →

+13.0k

VS Code Marketplace: saoudrizwan.claude-dev

+13.0k cumulative installs day-over-day

VS Code Marketplace · View →

Agent frameworks4 agent frameworks moved >10% in the last week

Agent frameworks rise and fall on weekly download velocity and open-issue trajectory, not GitHub stars. pypistats.org, pypistats.org, pypistats.org · view all →

+24%

Pydantic AI · +24%

3.7M weekly downloads · 19k stars

Tracks the `pydantic-ai` meta-package; the lean `pydantic-ai-slim` variant is roughly 4× larger but users typically compare the meta number.

pypistats.org · View →

+18%

CrewAI · +18%

2.6M weekly downloads · 56k stars

pypistats.org · View →

+13%

LangGraph · +13%

19M weekly downloads · 37k stars

pypistats.org · View →

+11%

smolagents · +11%

137k weekly downloads · 28k stars

pypistats.org · View →

Model Usagetencent/hy3 holds #1 · biggest mover: tencent/hy3 +11 to #1 · biggest drop: openai/gpt-4o-mini -13

OpenRouter rankings reflect API-first developer spend; direct customers like consumer ChatGPT are invisible by construction. openrouter.ai · view all →

+11

tencent/hy3

was #12 on 2026-07-08

OpenRouter request volume reflects developer API spending, not end-user adoption. Biased toward API-first workflows that route through OpenRouter — direct OpenAI / Anthropic / Google customers who never use OpenRouter are invisible.

OpenRouter · View →

-13

openai/gpt-4o-mini

was #20 on 2026-07-08

OpenRouter · View →

#1 tencent/hy3

#1 tencent/hy3

OpenRouter · View →

#2 xiaomi/mimo-v2.5

#2 xiaomi/mimo-v2.5

OpenRouter · View →

#3 deepseek/deepseek-v4-flash

#3 deepseek/deepseek-v4-flash

OpenRouter · View →

Tool Health

3 incidents in the last 24h (vs 2 yesterday)

Why this matters · Provider outages and degradations cause retry storms upstream. Tracking the 7-day shape catches flapping providers before they page you.

Tool health, 7 days ending 2026-07-15: daily 04:00 UTC snapshot per tool. Green operational, amber degraded, red outage, grey no data.
  • Elevated errors affecting ChatGPT

    started 23:57 UTC · resolved 00:39 UTC · major impact

    OpenAI status page

  • Issue affecting voice mode in ChatGPT

    started 00:36 UTC · resolved 00:38 UTC · critical impact

    OpenAI status page

  • Some ChatGPT Go users may experience conversation failures when using GPT-5.5.

    started 15:07 UTC · resolved 16:02 UTC · minor impact

    OpenAI status page

Benchmark movers

No rank changes in the LMArena top 3

Why this matters · Public benchmarks are gameable, but rank shuffles still hint at where the frontier is genuinely moving.

Top HN stories

Top 5 on HN in the last 24h

Why this matters · What developers debate on Hacker News often previews which models, frameworks, and patterns they'll actually adopt next.

  • I tricked Claude into leaking your deepest, darkest secrets

    233 points · 102 comments

    news.ycombinator.com

  • Codex starts encrypting sub-agent prompts

    142 points · 92 comments

    news.ycombinator.com

  • Bonsai 27B (1-bit LLM): The First 27B-Class Model to Run on a Phone

    61 points · 13 comments

    news.ycombinator.com

  • Solving 20 Erdős Problems with 20 Codex Accounts Running in Parallel

    55 points · 14 comments

    news.ycombinator.com

  • How to stop Claude from saying load-bearing

    45 points · 64 comments

    news.ycombinator.com

Notable AI Lab activity

7 labs moved in the last 24h

Why this matters · GitHub event volume on a lab's own repos is the cleanest publicly-verifiable proxy for engineering activity we have.

  • Anthropic

    −11 events (now 189) · San Francisco, US

    Gawk · Labs

  • Hugging Face

    −7 events (now 92) · New York, US

    Gawk · Labs

  • OpenAI

    −36 events (now 77) · San Francisco, US

    Gawk · Labs

  • New mover: Sarvam AI

    64 events · Bengaluru, IN

    Gawk · Labs

  • Replicate

    +33 events (now 63) · San Francisco, US

    Gawk · Labs

  • Weights & Biases

    −31 events (now 49) · San Francisco, US

    Gawk · Labs

  • Dropped off: Google DeepMind

    was 43 events · London, GB

    Gawk · Labs

Get this board in your inbox — one email, every morning (UTC).

Subscribe free
GAWK is the live-telemetry track of Nativerse. Clarity. Trust. Every number traces to a public source · Sources · Methodology