Live telemetry · daily brief
GAWK
by nativerse
● ARCHIVE
2026-05-21
2 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.
- ■ai downloads grew for the 7th consecutive snapshot.
- ■openai downloads grew for the 7th consecutive snapshot.
SDK adoption5 notable shifts across the six registries
Package install volume is where developers place real bets — distinct from where labs are shipping marketing. npmjs.com, crates.io, hub.docker.com · view all →
+329.4k
Docker Hub: ollama/ollama
+329.4k all-time pulls day-over-day
+16.0k
VS Code Marketplace: GitHub.copilot
+16.0k cumulative installs day-over-day
Agent frameworks2 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 · view all →
+51%
CrewAI · +51%
4.0M weekly downloads · 52k stars
-31%
Pydantic AI · -31%
8.1M weekly downloads · 17k stars
Tracks the `pydantic-ai` meta-package; the lean `pydantic-ai-slim` variant is roughly 4× larger but users typically compare the meta number.
Model Usagetencent/hy3-preview holds #1 · biggest mover: z-ai/glm-4.7 +11 to #30 · biggest drop: deepseek/deepseek-v4-flash -43
OpenRouter rankings reflect API-first developer spend; direct customers like consumer ChatGPT are invisible by construction. openrouter.ai · view all →
+11
z-ai/glm-4.7
was #41 on 2026-05-14
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.
-43
deepseek/deepseek-v4-flash
was #5 on 2026-05-14
—
#1 tencent/hy3-preview
#1 tencent/hy3-preview
—
#2 deepseek/deepseek-v4-flash
#2 deepseek/deepseek-v4-flash
—
#3 anthropic/claude-sonnet-4.6
#3 anthropic/claude-sonnet-4.6
Tool Health
2 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 errors on Claude Opus 4.7
started 14:43 UTC · resolved 16:04 UTC · minor impact
SWE-1.6 Fast Elevated Error rate
started 04:00 UTC · resolved 04:00 UTC · minor impact
Benchmark movers
2 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.
New to top 3: gemini-3.5-flash
#3 · google · 1486 Elo
Dropped from top 3: claude-opus-4-7-thinking
was #3 · anthropic
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.
An OpenAI model has disproved a central conjecture in discrete geometry
303 points · 198 comments
OpenAI to confidentially file for IPO as soon as Friday
52 points · 7 comments
Qwen3.7-Max: The Agent Frontier
35 points · 9 comments
PopuLoRA: Co-Evolving LLM Populations for Reasoning Self- Play
22 points · 2 comments
Mind-Blowing Growth Is About to Propel Anthropic into First Profitable Quarter
14 points · 3 comments
Notable AI Lab activity
9 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.
LangChain
+25 events (now 196) · San Francisco, US
Hugging Face
+13 events (now 106) · New York, US
Microsoft Research
−19 events (now 99) · Redmond, US
Ollama
+6 events (now 95) · San Francisco, US
Weights & Biases
+24 events (now 80) · San Francisco, US
Google Research
−20 events (now 72) · Mountain View, US
New mover: Replicate
52 events · San Francisco, US
DeepSeek
−31 events (now 51) · Hangzhou, CN
Dropped off: AMD
was 66 events · Santa Clara, US
Get this board in your inbox — one email, every morning (UTC).
Subscribe free