Gawk · 2026-05-15
Gawk — 2026-05-15 · 3 tool incidents
Five verifiable things that 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 24th consecutive snapshot.
- · diffusers downloads declined for the 4th consecutive snapshot.
- · torch downloads declined for the 3rd consecutive snapshot.
Benchmark movers
#3 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.
claude-opus-4-6
+2 Elo (now 1498) · anthropic
New to top 3: claude-opus-4-7-thinking
#3 · anthropic · 1485 Elo
Dropped from top 3: gemini-3.1-pro-preview
was #3 · google
Source: lmarena.ai
Tool Health
#3 incidents in the last 24h (vs 3 yesterday)
Why this matters · Provider outages and degradations cause retry storms upstream. Tracking the 7-day shape catches flapping providers before they page you.
Elevated error rates on requests to some models
started 00:18 UTC · resolved 01:46 UTC · major impact
Elevated errors on Claude Opus 4.7
started 20:33 UTC · resolved 22:25 UTC · minor impact
Codex Cloud and Code Review experiencing high failure rate
started 20:58 UTC · resolved 21:28 UTC
Source: status.anthropic.com, status.openai.com
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.
How Claude Code works in large codebases
147 points · 92 comments
Claude for Legal
53 points · 52 comments
Work with Codex from Anywhere
41 points · 13 comments
Codex is now available on mobile via ChatGPT app
26 points · 9 comments
The era of 15GB free Gmail storage is ending
17 points · 17 comments
Source: news.ycombinator.com
SDK adoption
#5 notable shifts across the six registries
Why this matters · Package install volume is where developers place real bets — distinct from where labs are shipping marketing.
PyPI: anthropic
+390.3k 24h downloads day-over-day
pypistats.org is a third-party aggregator — counts include mirror hits, CI builds, and pip install retries.
crates.io: ort
+26.8k 90d downloads day-over-day
Docker Hub: ollama/ollama
+363.1k all-time pulls day-over-day
Homebrew: ollama
−481 30d downloads day-over-day
VS Code Marketplace: GitHub.copilot
+23.6k cumulative installs day-over-day
Source: pypistats.org, crates.io, hub.docker.com, formulae.brew.sh, marketplace.visualstudio.com
Agent frameworks
#5 agent frameworks moved >10% in the last week
Why this matters · Agent frameworks rise and fall on weekly download velocity and open-issue trajectory, not GitHub stars.
CrewAI · +46%
2.9M weekly downloads · 51k stars
smolagents · +31%
156k weekly downloads · 27k stars
Pydantic AI · +18%
12M weekly downloads · 17k stars
pypistats.org · View on Gawk →
Tracks the `pydantic-ai` meta-package; the lean `pydantic-ai-slim` variant is roughly 4× larger but users typically compare the meta number.
AutoGen · +11%
369k weekly downloads · 58k stars
pypistats.org · View on Gawk →
Tracks the live `autogen-agentchat` package — the deprecated bare `autogen` PyPI name is a different abandoned project and is not counted here.
LangGraph · +11%
15M weekly downloads · 32k stars
Source: pypistats.org, pypistats.org, pypistats.org, pypistats.org, pypistats.org
Notable AI Lab activity
#13 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
+10 events (now 200) · San Francisco, US
New mover: LangChain
182 events · San Francisco, US
Microsoft Research
+12 events (now 107) · Redmond, US
OpenAI
−9 events (now 101) · San Francisco, US
Hugging Face
+20 events (now 98) · New York, US
DeepSeek
+35 events (now 96) · Hangzhou, CN
New mover: Ollama
96 events · San Francisco, US
New mover: Weights & Biases
89 events · San Francisco, US
New mover: Replicate
31 events · San Francisco, US
Dropped off: Stanford CRFM
was 30 events · Stanford, US
Dropped off: Google DeepMind
was 18 events · London, GB
Dropped off: INRIA
was 14 events · Paris, FR
Dropped off: Moonshot AI
was 13 events · Beijing, CN
Source: gawk.dev
Model Usage
#tencent/hy3-preview holds #1 · biggest mover: openrouter/owl-alpha +12 to #10 · biggest drop: qwen/qwen3.5-flash-02-23 -16
Why this matters · OpenRouter rankings reflect API-first developer spend; direct customers like consumer ChatGPT are invisible by construction.
openrouter/owl-alpha climbed +12 to #10
was #22 on 2026-05-08
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.
qwen/qwen3.5-flash-02-23 slipped -16 to #44
was #28 on 2026-05-08
#1 tencent/hy3-preview
#2 anthropic/claude-opus-4.7
#3 anthropic/claude-sonnet-4.6
Source: openrouter.ai