Live telemetry · daily brief
GAWK
by nativerse
● ARCHIVE
2026-08-21
4 tool incidents
What verifiably moved in the AI ecosystem in the last 24h. Every number traces to a public source.
What moved
- ■@anthropic-ai/sdk downloads grew for the 3rd consecutive snapshot.
- ■@langchain/core downloads grew for the 3rd 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 →
+362.6k
Docker Hub: ollama/ollama
+362.6k all-time pulls day-over-day
+10.7k
VS Code Marketplace: saoudrizwan.claude-dev
+10.7k 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 →
+99%
CrewAI · +99%
12M weekly downloads · 57k stars
-18%
smolagents · -18%
139k weekly downloads · 29k stars
Model Usagedeepseek/deepseek-v4-flash-0731 holds #1 · biggest mover: deepseek/deepseek-v4-pro-0813 +28 to #15 · biggest drop: google/gemma-4-31b-it -12
OpenRouter rankings reflect API-first developer spend; direct customers like consumer ChatGPT are invisible by construction. openrouter.ai · view all →
+28
deepseek/deepseek-v4-pro-0813
was #43 on 2026-08-14
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.
-12
google/gemma-4-31b-it
was #23 on 2026-08-14
—
#1 deepseek/deepseek-v4-flash-0731
#1 deepseek/deepseek-v4-flash-0731
—
#2 tencent/hy3
#2 tencent/hy3
—
#3 xiaomi/mimo-v2.5
#3 xiaomi/mimo-v2.5
Tool Health
4 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.
Elevated errors on requests to multiple models
started 19:16 UTC · resolved 19:42 UTC · major impact
Elevated errors on Google connectors
started 18:32 UTC · resolved 19:01 UTC · minor impact
Elevated Codex API authentication errors
started 16:58 UTC · resolved 17:52 UTC · minor impact
Elevated Errors for Thinking mode in ChatGPT
started 13:56 UTC · resolved 16:51 UTC · minor impact
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.
#1 claude-opus-5-high
anthropic · 1505 Elo
#2 claude-opus-5-max
anthropic · 1503 Elo
#3 claude-opus-4-6-high
anthropic · 1503 Elo
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.
Codex on AWS bedrock bug causing 10x charges
107 points · 31 comments
Clean up Claude 5's token vomit with a separate LLM
53 points · 38 comments
I Am Morally Opposed to Updating My Claude.md
27 points · 21 comments
Hacking with Claude on a $27 Smart Watch
27 points · 10 comments
DiffusionGemma Technical Report
27 points · 4 comments
Notable AI Lab activity
10 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
+10 events (now 200) · San Francisco, US
Anthropic
−22 events (now 172) · San Francisco, US
OpenAI
+5 events (now 143) · San Francisco, US
Weights & Biases
−8 events (now 85) · San Francisco, US
Hugging Face
−22 events (now 76) · New York, US
Microsoft Research
−17 events (now 57) · Redmond, US
DeepSeek
−6 events (now 54) · Hangzhou, CN
AMD
+9 events (now 33) · Santa Clara, US
New mover: Meta FAIR
29 events · Menlo Park, US
Dropped off: Replicate
was 26 events · San Francisco, US
Get this board in your inbox — one email, every morning (UTC).
Subscribe free