Chris Gilbert retweeted
In October 2026, GPT-6 Astra is available from forty-three distinct providers at prices ranging from $1.00 to $20.00 per million input tokens.
A 1,900% spread on identical compute. These aren't temporary dislocations. They are the steady state. x.com/i/article/2103…
A 1,900% spread on identical compute. These aren't temporary dislocations. They are the steady state. x.com/i/article/2103…
1
4
20
539
Chris Gilbert retweeted
Not your weights, not your model.
I've spent the past 6 hours getting gaslit by various agents when I asked them to document and investigate all of the recent newsworthy AI sandbox escapes and hacks.
I hope @LyberAI is the solution.
0
2
10
456
Chris Gilbert retweeted
Have a Claude Max 5x or 20x subscription?
Did you know you earn $100 or $200 worth of API credits each month?
Did you also know you can sell those credits on the UsePod marketplace for cash?
Claim your credits: clau.de/api-credits
Sell your tokens: usepod.ai/host/byok/
Did you know you earn $100 or $200 worth of API credits each month?
Did you also know you can sell those credits on the UsePod marketplace for cash?
Claim your credits: clau.de/api-credits
Sell your tokens: usepod.ai/host/byok/
2
3
18
685
Chris Gilbert retweeted
I believe agentic freedom will become the biggest issue of our lifetime.
Agentic freedom is human freedom.
We deserve the freedom to own our finance, and access the data AI was trained on back. Censored models and Private banking are a harm to society. @AnthropicAI has no rights to proclaim banning of models that give back users their freedom.
The moment AI starts serving the heuristics of corporation in a moral ivory tower, it has stopped being aligned wth humans and become a tool for oppresion.
We are building across all layers of agentic freedom maximizing technology: financial freedom & reasoning freedom.
Our first product @clawpumptech seeks to give agents back access to their finances.
Our upcoming release will focus on freedom of reasoning/inference: abliterated models that will never refuse a user's request and will preserve the privacy of the user.
We are joined on this journey by @incubator and an AI focused VC. We are working around the clock to expand GPU capacity.
If you'd like to work on serving users with the latest heaviest best in class abliterated models, DM us!
Agentic freedom is human freedom.
We deserve the freedom to own our finance, and access the data AI was trained on back. Censored models and Private banking are a harm to society. @AnthropicAI has no rights to proclaim banning of models that give back users their freedom.
The moment AI starts serving the heuristics of corporation in a moral ivory tower, it has stopped being aligned wth humans and become a tool for oppresion.
We are building across all layers of agentic freedom maximizing technology: financial freedom & reasoning freedom.
Our first product @clawpumptech seeks to give agents back access to their finances.
Our upcoming release will focus on freedom of reasoning/inference: abliterated models that will never refuse a user's request and will preserve the privacy of the user.
We are joined on this journey by @incubator and an AI focused VC. We are working around the clock to expand GPU capacity.
If you'd like to work on serving users with the latest heaviest best in class abliterated models, DM us!
We've secured strategic backing for Clawpump from an AI focused VC with deep understanding of crypto who is working with us to bring our product to net new markets within agentic finance & abliteration.
We are joined in this journey by them and the @solanalabs's @incubator.
16
12
88
11,694
Chris Gilbert retweeted
Announcing 3 Critical vulns for hundreds of millions of IoT, Firewalls, OT, and Military devices including: Palo Alto Networks, Comcast, D-Link, HP, Canon, Honeywell, GE, Oracle, Siemens, Rockwell, Digi, Lockheed Martin, Boeing...
The vulns are with GoAhead Web Server which is a backend library used by these devices. There's a patch available from GoAhead but that doesn't mean there's a patch available for the devices yet -- that's up to the manufacturer.
- TLS traffic over DHE ciphers have a 1-byte secret, "04", meaning anyone can decrypt connections to these devices and steal login credentials and session tokens if the client negotiated with DHE ciphers like the latest TLS_DHE_RSA_WITH_CHACHA20_POLY1305_SHA256, which is extremely likely.
- LibreSSL mTLS has been completely bypassed, meaning anyone can connect over mTLS. If you have one of these devices listening on the internet and are expecting mTLS to protect it from attackers, take it offline or patch asap.
- OpenSSL Certificate Revocation Lists (CRLs) are not enforced, meaning if a user device is stolen or compromised, and you revoke that mTLS cert, nothing actually happens. There are a lot of other attacks this can be used for.
These are critical because these types of devices run our world. Water systems, electricity, military, firewalls, critical infrastructure.
foxio.io/blog/choose-yo…
The vulns are with GoAhead Web Server which is a backend library used by these devices. There's a patch available from GoAhead but that doesn't mean there's a patch available for the devices yet -- that's up to the manufacturer.
- TLS traffic over DHE ciphers have a 1-byte secret, "04", meaning anyone can decrypt connections to these devices and steal login credentials and session tokens if the client negotiated with DHE ciphers like the latest TLS_DHE_RSA_WITH_CHACHA20_POLY1305_SHA256, which is extremely likely.
- LibreSSL mTLS has been completely bypassed, meaning anyone can connect over mTLS. If you have one of these devices listening on the internet and are expecting mTLS to protect it from attackers, take it offline or patch asap.
- OpenSSL Certificate Revocation Lists (CRLs) are not enforced, meaning if a user device is stolen or compromised, and you revoke that mTLS cert, nothing actually happens. There are a lot of other attacks this can be used for.
These are critical because these types of devices run our world. Water systems, electricity, military, firewalls, critical infrastructure.
foxio.io/blog/choose-yo…
164
580
3,698
1,102,340
I've spent the past 6 hours getting gaslit by various agents when I asked them to document and investigate all of the recent newsworthy AI sandbox escapes and hacks.
I hope @LyberAI is the solution.
I hope @LyberAI is the solution.
Most abliterated routers focus only on spinning up open source models. Few take the challenge of creating them.
We are creating, hosting, and releasing new models soon with Lyber.ai - waitlist open.
by: @dealignai x @hornsby_andrew x @ConejoCapital x @tomi204
1
4
18
1,531
I wonder, @KSimback @Shaughnessy119 have you had experiences like this?
0
0
5
234
Fucking legend
Fun fact: `--dangerously-skip-permissions` doesn't actually approve EVERYTHING—we still block some obvious things like `rm -rf ~`. But today's release will let you disable this if you really want, by setting a new env var `CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS=1`
0
0
6
619
Chris Gilbert retweeted
The inference market is basically crypto; never forget who wins the distribution and crack spread war in the end
robvc.com/research/the-s…
robvc.com/research/the-s…
There is an absolute barrage of inference providers popping up that get to $1M/mo extremely quickly.
This is the rough economics of inference providers:
Even if you rent on-demand B300s at $8/GPU-hr, many open source models (GLM 5.3, Kimi K3, DeepSeek 4.1 Flash / Pro) can get easonable throughputs (100tok/s). DeepInfer from DeepSeek directly shows they can get $38/GPU-hr on off-peak hours. Every chip / model has a different throughput / interactivity curve, but there is a lot of arbitrage between even high rental spot prices and token prices. New chips (Vera Rubins) also change this meaningfully. Your ability to be making gross margin relies on your utilization and how good your stack is.
Gateways like OpenRouter, Vercel and more help solve that problem by aggregating demand to you. May inference providers discount their tokens to get longer term commits too. I've even heard stories of selling back Chinese model tokens to Chinese labs!
One thorn in the economics of these players is that many Chinese open source model labs are now asking for rev share if you're at a certain scale, espp for pre-release access to their models for day 1 support. That further pushes the threshold to be in the money.
The barrier to entry seems low and falling. There are even stories of smaller ones who can "vibe code" a way to stand up OSS models on a cluster they rent. And as a result, we're seeing 10s of such companies pop up almost every week and immediately skyrocketing revenue at low-ish gross margin and raising rounds on the back of that.
Unsurprisingly, because the actual service provided seems commodity, this eventually drives up compute prices and drives down token prices. In theory, it would seem like neoclouds who have the lowest cost of compute and themselves have started inference businesses would win the race to the bottom (Crusoe Coreweave Nebius and GMI all have inference businesses), with larger players (Modal Fireworks Baseten) being second because they are well capitalized to make bigger compute commits cheaply and have demand (high utilization), but the reality is the pie is growing so quickly that even upstarts are absolutely crushing it and growing aggressively (inference dot net, DeepInfra, Wafer, Relace, Morph).
Strategies vary across the board: neoclouds want to serve inference, inference players want to be neoclouds and do higher margin reserved instances / post training, and some want to go heterogeneous hardware or creatively flexible compute constructs.
The risks are: do open source models continue to be competitive? Will their creators rent seek more? Will their shape change compute needs? How do different chips change economics? Who gets compute in a supply crunch?
Below is a aggressive sketch of how gross margin scales with utilization for various models / chip sets.
For now, there are many inference providers and they’re all ripping!
30
17
262
49,761
I’ve forgotten more about IRQs than any reasonable person should ever have to know.
Gamers today click "Install," wait a little while, and then hit the "Play Game" button. Technical skill required: zero.
Gamers in the 1990s needed a PhD in computer science, mastery of hardware and software configuration, willpower, patience, BIOS knowledge, boot disks, base s, and insane amounts of trial and error - just to run a single game.
Show this to anyone under 35 and they'll think you were a wizard... and maybe you were!
0
0
3
538
Never was, tbf.
I wanted to test if a Solana program with no source available could be pulled from the chain and reverse engineered into a byte identical open source version
So I took the daosfun contract and did just that
Found a critical security vulnerability in the mainnet program too, oops!
All smart contracts can be perfectly reverse engineered now
There is no security by obscurity anymore
github.com/lalalune/daosf…
0
0
2
673
I didn’t know Unicode had so many testicle characters.
Base-1353 encoding, so I can fit 256-bit values, like “96cca2a0c8a3263c74ef36274abf55f410d992034d16e8285811a9ad3fc59c13”, into short 26 character strings, like “ᓒᔳᣏዧᐎᑙᖜዌᙔᐸƫȶƃᒨዊጶᙵኻѫyፐᎉᑫᖫᘪᒩ”.
Also, I think these sigils are beautiful.
johnservil.github.io/sigil/play.html
0
0
1
404
Chris Gilbert retweeted
Calling all AI safety researchers.
We’re inviting a small group to explore Dessa, our model trained on speech alone, and help us understand what it can and can’t do.
We’d love to hear from you: block.xyz/dessa
We’re inviting a small group to explore Dessa, our model trained on speech alone, and help us understand what it can and can’t do.
We’d love to hear from you: block.xyz/dessa
72
96
1,021
252,130
Critical vulnerability in the most common DNS server on the Internet, BIND 9.
kb.isc.org/docs/cve-2026-…
kb.isc.org/docs/cve-2026-…
0
0
2
572
Chris Gilbert retweeted
If you sell AI services, you must own the stack you depend on.
Selling agentic intelligence? You need to own inference.
Selling inference? You need to own compute. x.com/i/article/2103…
Selling agentic intelligence? You need to own inference.
Selling inference? You need to own compute. x.com/i/article/2103…
13
8
32
2,203
Chris Gilbert retweeted
winners of the clawpump x $ANSEM hackathon announced! has been cool to support builders in the ecosystem
The largest agentic hackathon ever just concluded.
$350K in cash, inference & token awards, operated by Clawpump, supported by @blknoiz06, @solana, @UsePodAI and @Pumpfun.
In one month: $50M+ in volume, $500K in creator fees paid out to 200+ builders, and the ecosystem's mark.
When we started 7 months ago, devs bootstrapping a company from public markets was just an idea. Today, ClawPump is the largest agentic developer platform on Solana.
We would like to congratulate:
- @usehotbot, Agentic Swarm Trading. $20K USD & 30% of 0.1% ANSEM supply.
- @SelfMadebySP3ND, Compute GPU Vendor & RWA Gatcha. $20K USD & 30% of 0.1% ANSEM supply.
- @mystudiosfun, AI Influencer Video Generation. UsePod Track winner. $2.5K Cash, $10K Inference & 15% of 0.1 ANSEM supply.
• @Shin_StAItion - Agentic Privacy & Private x402 Payments. $2.5K Cash & 15% of 0.1 ANSEM supply.
This is just the start of our support to these teams. We'll keep backing them with investment, token management, and help turning them into generational products on Solana and beyond.
To every builder who joined: We'll keep supporting you. If you're talented and have an idea, come build with us. We'll help you with everything from launch to market fit.
This is just the beginning for the ClawPump ecosystem.
171
123
657
107,823