Permanent copy of issue 40. Today's issue is always the front page.
OpenAI said an internal model proved that the Navier-Stokes equations can blow up, meaning a fluid's speed becomes infinite at some points. Scientific American reported the claim, and mathematician Tristan Buckmaster alleged the lab used a method he had been developing; OpenAI denied that. Google DeepMind released AlphaGenome Atlas, a 1-petabyte map of predicted effects for about 9 billion single-letter DNA changes. CISA, NSA and the FBI said China-based AI firms extracted billions of tokens from US frontier models through industrial-scale distillation.
Date: 2026-09-09
Updated: 2026-09-09 06:06
Status: LIVE
Broadcast
Sebastien Bubeck, who leads OpenAI's math team, said the internal model solved the related Euler problem by a different route than Tristan Buckmaster and Levent Alpoge, and that the lab did not use their prompts. The Clay Mathematics Institute still offers $1 million for a Navier-Stokes resolution; some mathematicians say the forcing method leans on a term many experts treat as optional. AlphaGenome Atlas is live as a no-code website portal plus an API. Meta said Muse runs inside a dedicated secure virtual machine with its own browser.
Watch
Stories
OpenAI said an internal model proved that the Navier-Stokes equations can blow up, meaning they dictate that a fluid's speed becomes infinite at some points. Scientific American reported the claim; mathematician Tristan Buckmaster alleged the lab used a method he had been developing, and OpenAI denied that.
-- A Millennium Prize claim from a lab model is cluster research, not a desk proof.
2026-09-08 / Scientific American / chip, efficiency / central / #openai-claims-navier-stokes-solution
Google DeepMind released AlphaGenome Atlas, a database that predicts the effects of every possible single-letter DNA change in the human genome. The lab said it precomputed about 9 billion variants into a 1-petabyte dataset and added an AlphaGenome Variant Impact score for coding and non-coding regions.
-- A petabyte genome map is a cluster product. Researchers query it; they do not train it at home.
2026-09-08 / Google / chip, efficiency / central / #deepmind-maps-9-billion-dna-variants
CISA, the NSA and the FBI said China-based AI companies extracted billions of tokens from US frontier models through industrial-scale knowledge distillation. The September 8 advisory named DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun and Z.AI, and said the campaigns targeted variants of Claude, GPT, Gemini and Grok since at least late 2024.
-- Open-weight labs are being accused of renting US APIs to skip training cost. That is not a local runtime.
2026-09-08 / CISA / efficiency, chip / mixed / #us-agencies-warn-of-model-distillation
Meta introduced Muse, a personal AI agent that the company said will take on tasks rather than only answer questions. Muse is rolling out in the US on iOS, Android and muse.ai, running inside what Meta calls a Muse Secure VM, with a free tier and paid plans.
-- A personal agent on a Meta VM still meters work in the cluster.
2026-09-08 / Meta / chip / central / #meta-launches-muse-personal-agent
Axis
local [####################----------] central
Score: 67
clusters still winning
up 1
OpenAI says an internal model proved Navier-Stokes can blow up. DeepMind shipped a 1-petabyte genome atlas. US agencies say China-based labs distilled US APIs at industrial scale. The work still sits in the cluster.
Forces
| Force | Meter | Score | Pull | Note | ||
| CHIP / CAPEX |
|
84 | central | OpenAI's Navier-Stokes claim came from an internal model, not a desk. DeepMind precomputed 9 billion DNA variants into a 1-petabyte atlas. | ||
| ENERGY |
|
87 | central | A petabyte genome map and a weekend of cluster math still burn central power. No new gigawatt lease today, but the jobs are not local. | ||
| EFFICIENCY |
|
45 | local | US agencies say distillation of Claude, GPT, Gemini and Grok is how some labs skip research cost. That is API extraction, not a desk runtime. |
Line charts (0=local 100=central 2026-08-24 to 2026-09-09)
Market overlay (2026-08-11 to 2026-09-09)
Latest: BTC $79,008 / hashrate ~929.89 EH/s / token markers avg $5.3033/M / axis 67. These lines are different units. We compare timing, not dollars-to-hashrate.
Not shown: GPU cloud $/hour, hub electricity, miner hashprice, or lab cost per token -- none of those are a clean public feed.
Power and list prices
BTC $79,008 / hashrate ~930 EH/s. List-price gap (not a subsidy %): 44.4x list gap (blended) GPT-5.6 Luna -> Claude Fable 5 ($0.45/M to $20.00/M). (mempool.space + Coinbase/CoinGecko)
Marker prices (USD per 1M tokens; verified 2026-08-19; blended = (3*input + 1*output) / 4; -- = open/self-host or no single public list)
| # | Model | Org | CC | Weights | Tier | In | Out | Blend |
| 1 | Claude Sonnet 5 | Anthropic | US | closed undisclosed |
workhorse | $2.00 | $10.00 | $4.00 |
| 2 | GPT-5.5 | OpenAI | US | closed undisclosed |
flagship | $5.00 | $30.00 | $11.25 |
| 3 | DeepSeek V4 Flash | DeepSeek | CN | open ~284B total / ~13B active (MoE) |
commodity | $0.44 | $1.32 | $0.66 |
If useful, support the broadcast.
bc1qfs3yw5qlq8sxs50crzh3g84ug27gvww9npu85n
Subscribe
$ echo you@email.com | underclassic --subscribe
Previous issue | This week | Next issue | archive
One email per issue. No tracking pixels, no filler.
home | rss | page 2 | archive | models | subscribe | data | board | mcp | llms.txt