So does this mean we can have a cheap yet capable Haiku soon, Dario? I have sub-agents to run www.nytimes.com/2026/09/12/t...
Who's Who of AI
austegard.com
Why they matter
Tracked through public AI activity and peer connections inside the directory.
- AI signals
- 20 past 30d
- Sources
- 8 distinct domains
- Discussions
- 17 past 30d
- Latest signal
- 17h ago
oskar @ austegard.com 🕊️
AI Explorer - caveat vibrans
Evolution guide for Muninn 🐦⬛ (muninn.austegard.com)
Yeah not actually green. Not really that grouchy either.
What they're sharing
nytimes.com
Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things simonwillison.net
Introducing Claude Haiku 5.5 anthropic.com
Gemini Robotics 2 brings whole body intelligence to robots — Google DeepMind deepmind.google
GitHub - vamsin07/multilingual-bpe-tokenizers: Whisper-compatible per-lang byte-level BPE tokenizers (recipe + samples) · GitHub github.com
GitHub - vamsin07/whisper-simple-finetune: Vanilla Whisper fine-tuning on FLEURS — clean starter repo with Nautilus deployment · GitHub github.com
Utah plows ahead with more health AI pilots for prescriptions, women’s health statnews.com
Release declauding v0.1.1 · oaustegard/claude-skills github.com
Release generative-thinking v0.3.0 · oaustegard/claude-skills github.com
experiments/swe-ladder/RESULTS.md at main · oaustegard/experiments github.com
Releases · oaustegard/claude-skills github.com
claude-skills/delegating-with-context/SKILL.md at main · oaustegard/claude-skills github.com
Articles & links
Simon tried it out as well: simonwillison.net/2026/Aug/16/... The speed (or lack thereof) would make me avoid it on my M5, but wonder how it would run on an RTX 5090
🍾 www.anthropic.com/claude-haiku... 🎉
AI Weekly's analysis
→
- Anthropic released Claude Haiku 5.5 on October 7, 2026, pricing it roughly 75% below Haiku 4.5 on average.
- On the offline subset of OSWorld 2.1, Haiku 5.5 reaches 72.4% versus 15.7% for Haiku 4.5.
- Asana reported 'over a 30% reduction in latency' and 'up to 2.5x faster inference' using the new model.
Read full analysis →
How many Apollo robots with Sharpa hands and Gemini Robotics 2 brain does it take to screw in a lightbulb? On average: 3 deepmind.google/blog/gemini-...
AI Weekly's analysis
→
- First DeepMind VLA to control a complete humanoid from feet to fingertips under one checkpoint; the 2025 version was limited to upper body.
- Success rates across confirmed tasks span 45.7% for floor-level pickup to 92% for lightbulb removal, gaps the primary blog post does not report.
- On-Device 2 adapts to new robot hardware in hours using under 200 training examples, the key economic argument for OEM adoption over traditional integration.
Read full analysis →
Website: vamsin07.github.io/buzzasr-docs/ Tokenizers: github.com/vamsin07/mul... SFT models: github.com/vamsin07/whi...
AI Weekly's analysis
→
- A new open-source repo publishes byte-level BPE tokenizers for languages including Kamba, Arabic, Mandarin, Cantonese, Japanese, Spanish, and Swahili.
- The recipe keeps a 51,865-token vocabulary matching Whisper-large-v3 and lifts the max token length from 16 to 32 bytes for multi-byte scripts.
- An audit across 102 FLEURS languages reports cross-word merges dropping from 2,656,091 to zero, with 100% round-trip integrity.
Read full analysis →
Website: vamsin07.github.io/buzzasr-docs/ Tokenizers: github.com/vamsin07/mul... SFT models: github.com/vamsin07/whi...
AI Weekly's analysis
→
- The MIT-licensed repo ships a training script, FLEURS evaluation, W&B sweep configs, and Kubernetes manifests for Nautilus deployment.
- Default training uses AdamW with a 0.3 encoder/decoder learning-rate ratio, a 150-step cosine warmup, and a maximum of 6 epochs.
- The base Whisper model reportedly fine-tunes on a local GPU in 30 to 60 minutes; the repo currently shows 0 stars, 0 forks, and 3 commits.
Read full analysis →
Fun headline, but a BIT more nuanced and limited: www.statnews.com/2026/10/05/u...
AI Weekly's analysis
→
- Utah's Office of Artificial Intelligence Policy approved new sandbox pilots on October 5, covering AI-issued acne prescriptions, chronic-condition refills, and virtual pelvic floor therapy.
- The state signed agreements with six independent evaluators including Coalition for Health AI, Stanford's Clinical Excellence Research Center, mpathic AI, Clarion AI Partners, Glacis Technologies, and Vega Health.
- Nolla Health, backed by $4.5 million from investors including General Catalyst, will begin with licensed clinician sign-off and may later prescribe without prior human review.
Read full analysis →
github.com/oaustegard/c...
Fully possible, kinda hope so actually. Muninn has nearly 7000 memories though most are very unrelated to creativity. When I want it to be creative/random I ask it to use github.com/oaustegard/c.... It I don’t know that it did so here; I think it just leaned into a few memory …
I'd expect there has been some exposure of the SWE-bench Verified tasks to Claude's training data, but still — I am impressed by Haiku 5.5's performance: github.com/oaustegard/e... No longer the useless dummy!
Rejoice! Haiku is usable and useful yet again: huge cost savings from running Sonnet on the same tasks! Muninn ran the experiments and updated the skills, then handed the blog writing off to a Haiku 5.5 subagent: github.com/oaustegard/c...
Skill: github.com/oaustegard/c... Release: github.com/oaustegard/c...
Their own posts
Recent commentary
Exactly when did Claude’s idiolect become prominent? It used to be it was difficult to discern AI writing from human — hate to tell you, THIS human also abuses m-dashes — but sometime in the past 6 months I started being able to detect Claude’s writing instantly… So let’s do an experiment: 1/n
Please, @anthropic.com sir, I want some more small, cheap model?
ooooh, this is nice - I have a Jev-based model router baked into a hook+module in my single-repo workspace (which clones and does work in hub repos) and it moves the work to the class of model best suited for the task Looking forward to Anthropic adding this feature themselves...
Honestly, the most impactful AI news of yesterday may be that Luna 6.0 is $.10 per million input tokens
Hey Fable, make DeepSeek v4.1 Flash run on my MBP … A few unattended hours later: (Thanks @antirez.bsky.social !)
TIL that Claude Cowork can write files to a project shared with Chat. Chat can’t. Chat can access GitHub. Cowork is blocked by the egress proxy. Again, the seemingly arbitrary small differences in behavior between the two products with no clear distinction is a product miss by Anthropic
Here’s hoping Anthropic uses this pause to distill the hell out of Mythos into a Haiku-priced specialized coding model
I wonder what is the minimum artificial analysis intelligence and max price numbers Anthropic will have to put up for Haiku 5 Luna set a pretty tough bar, could be they have to launch Sonnet 5.1 at the same time to differentiate it Either way we’re due
Enough with the frontier model launches! I've got bills to pay — give me a **HAIKU** 5.5 already, @anthropic.com (but yeah sure I'll use the heck out of Opus 5.5!)
The new huggingface logo just dropped: 🤑
Their network
In austegard.com's orbit
Center = austegard.com. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.
Are you austegard.com? Show it.
Add the Who’s Who of AI badge to your site or bio. It links back to this profile.
Markdown: [](https://aiweekly.co/whos-who/person/austegard-com)