Founder @holoweave. I mostly post about agents and a bit of investing. Built product orgs at CoinMarketCap, Binance, Chainlink and Zendesk.
Joined April 2012
- Tweets589
- Following226
- Followers343
- Likes141
Rigario retweeted
Hi,
Tomorrow we are re-opening the Pro $200 subscriptions to new subscribers, but together with it we are also changing how we calculate the usage for it. In effect, if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan.
Now that it's said, let me explain why this is happening and why you will still get more work done than if you were on the Pro $200 subscription one month ago.
(a) We didn't want to compromise in other ways and are committing to not reintroducing the 5h limit, so that you can fully use the weekly usage when you want.
(b) On the subscription, we guarantee that over time you always get more work done and with an increasing level of quality. This means that you will continue to get more value per dollar spent as a result of models getting more efficient and us passing down the improvements in the form of API price reductions.
(c) We don't want to put an incentive on ourselves to artificially inflate the API list prices to make it look like you are getting a lot (and workaround it through discounts, etc). Instead we want to continue to both rapidly reduce prices and increase capabilities of models on the API. This week we introduced GPT-6 Sol and GPT-6 Luna at 50% of their previous price. Over time, we see prices go low enough that it makes sense for most to buy usage as needed without there being a significant gap between what you get in a subscription and what you get in the API for a dollar spent.
(d) Tomorrow, we are adding more things to the subscription that won't draw on the usage, I won't reveal what that is yet.
I wanted to be transparent before all the big announcements tomorrow. Lots of new exciting things are coming to the subscriptions that will make it super compelling, but I wanted to make sure to share this change ahead of time so you can all understand it before we shower you with good news.
Codexingly,
Tibo
Rigario retweeted
Grok 4.7 is here.
It's a notable improvement over Grok 4.6 at the same price and speed.
Rigario retweeted
Cormint re-affirms its commitment to be a good neighbor and partner at the city, county and state level.
We are grateful to be in Texas and applaud @GregAbbott_TX for his leadership in responsibly stewarding Texan economic growth.
Rigario retweeted
Couldn't have wrote a better blog in favor of Open Source AI or Hermes Agent's self improvement loop!
Thanks for the strong words in support of taking back the power to the individuals and businesses who really ought to own their stack.
Rigario retweeted
Fun interactive science app ideas | Part 3
Played around with generating 3D biological structures and made an app to explore them interactively
UI Design
GPT Images 2
Code
Gemini 3.1 Pro
More demos ↓
Rigario retweeted
// HeavySkill //
One of the cleaner takes on agentic harness design I've read.
They argue that what actually drives agent harness performance is not the orchestration code. It's a single inner skill: parallel reasoning followed by deliberation.
If you can internalize that into the model and most of the scaffolding becomes optional.
The paper systematizes this as a two-stage pipeline you can run beneath any harness, then trains it as a learnable skill via RLVR.
The numbers:
> GPT-OSS-20B jumps from 69.7% (M@K) to 85.5% (HM@4) on LiveCodeBench under the heavy-thinking variant.
> R1-Distill-Qwen-32B nearly doubles on IFEval, from 35.7% to 69.3%.
> Several models reach Pass@N-level performance with HeavySkill.
Harness wins start to look like model wins once you can train them in. If parallel-reasoning-plus-deliberation really is the inner skill, the long arc is models that come with it baked in, not orchestration glue around them.
Paper: arxiv.org/abs/2605.02396
Learn to build effective AI agents in our academy: academy.dair.ai/
Rigario retweeted
I pulled 3,182 tweets analyzing @NousResearch Hermes Agent versus Claude Code to understand exactly why developers are choosing one or the other
I want to figure out why people use Hermes vs Claude/ Claude Code
Here is the breakdown using 3,182 tweets to distill the answer vs guessing 🧵
Rigario retweeted
DeepSeek is back among the leading open weights models with the release of DeepSeek V4 Pro and V4 Flash, with V4 Pro second only to Kimi K2.6 on the Artificial Analysis Intelligence Index
@deepseek_ai has released DeepSeek V4 Pro and V4 Flash. V4 is the first new architecture from DeepSeek since V3. V4 introduces a new architecture with V4 Pro at 1.6T total / 49B active parameters and V4 Flash at 284B total / 13B active parameters, and is DeepSeek's first two-tier lineup, with Pro positioned for maximum capability and Flash for faster, lower-cost inference. Both models are hybrid thinking/non-thinking. We tested the reasoning variants in Max Effort and High Effort.
A year ago, DeepSeek R1 and R1 0528 were the leading open weights reasoning models on the Intelligence Index. Since then, several other open weights labs have released strong reasoning models, and V4 Pro now enters as the #2 open weights reasoning model on the Artificial Analysis Intelligence Index, behind only Kimi K2.6 (54). V4 Pro and V4 Flash remain also text input and output only.
Key takeaways:
➤ Large 10 point gain in Intelligence Index: DeepSeek V4 Pro (Max) scores 52 on the Artificial Analysis Intelligence Index, up from 42 for V3.2, making it the #2 open weights reasoning model behind Kimi K2.6. That said, if the weights to MiMo-V2.5-Pro are released as they have been for other Xiaomi models then it will place third. V4 Flash (Max) scores 47, behind V4 Pro but ahead of DeepSeek V3.2. This places it behind frontier models and at Claude Sonnet 4.6 (max) level intelligence.
➤ Leading agentic performance among open weights models: DeepSeek V4 Pro (Max) leads open weight models on agentic real-world work tasks, scoring 1554 on GDPval-AA. This places it ahead of Kimi K2.6 (1484), GLM-5.1 (1535), GLM-5 (1402), and MiniMax-M2.7 (1514).
➤ Gains in knowledge but an increase in hallucination rate: DeepSeek V4 Pro (Max) scores -10 on AA-Omniscience, an 11 point improvement over V3.2 (Reasoning, -21), driven primarily by higher accuracy. V4 Flash (Max) scores -23, broadly in line with V3.2. V4 Pro and V4 Flash both have a very high hallucination rate of 94% and 96% respectively meaning when they don’t know the answer they nearly always respond anyway.
➤ Flash materially behind Pro but well positioned for its size: DeepSeek V4 Flash (Max) scores 47 on the Artificial Analysis Intelligence Index, well below V4 Pro. However, at 284B parameters it is much smaller and is well positioned on the Intelligence vs Size frontier, sitting next to MiniMax-M2.7
➤ Cheaper than frontier models but more expensive than other open weights models and a large increase over DeepSeek V3.2: DeepSeek V4 Pro costs $1,071 to run the Artificial Analysis Intelligence Index. This makes it more than 4x cheaper than Claude Opus 4.7 ($4,811), but it remains more expensive than several other open weights models, including Kimi K2.6 ($948), GLM-5.1 ($544), DeepSeek V3.2 ($71), and gpt-oss-120B ($67). DeepSeek V4 Flash is much cheaper at $113.
➤ High token usage: DeepSeek V4 Pro uses 190M output tokens to run the Artificial Analysis Intelligence Index, making it one of the most token-intensive models tested. DeepSeek V4 Flash is even higher at 240M output tokens. This high token usage helps explain why V4 Pro’s total cost remains relatively high versus other open weights models despite low per-token pricing.
Key model details:
➤ Context window: 1M tokens, an 8x expansion on V3.2's 128K context window
➤ Modalities: Text input and output only, matching V3.2
➤ Size: DeepSeek V4 Pro 1.6T total / 49B active; V4 Flash 284B total / 13B active
➤ License: MIT
➤ Availability: Available on DeepSeek's first-party API; we expect many third-party providers to host the models
➤ Pricing: DeepSeek V4 Pro $1.74 / $3.48 per 1M input/output tokens; V4 Flash $0.14 / $0.28 per 1M input/output tokens. Cache hit input token pricing is $0.145 for V4 Pro and $0.028 for V4 Flash per 1M tokens. V4 Pro is significantly more expensive than past DeepSeek R1 and V3 models
Rigario retweeted
New subscription tiers are live on Nous Portal
→ Plus ($20)
→ Super ($100)
→ Ultra ($200)
Bonus credits on signups, upgrades, and renewals: +$2 on Plus / +$10 on Super / +$20 on Ultra
All tiers include access to:
→ 300+ models
→ Bundled tool usage
portal.nousresearch.com/mana…
Rigario retweeted
Step 3.5 Flash is now live for Nous Portal users, free for the next 10 days.
If you're running Hermes Agent with Nous Portal as your provider, run 'hermes update' and then 'hermes model' to configure access for Step 3.5 Flash (free).
Thank you @StepFun_ai for making this happen!
Rigario retweeted
The Hermes Agent Creative Hackathon starts now
16 Days, $25k in Prizes
Presented by @Kimi_Moonshot & @NousResearch
For the tinkerers pushing Hermes Agent into creative domains: video, image, audio, 3D, long-form writing, creative software, interactive media and more.
Show us what your Hermes Agent can do.
Details Below ↓
Rigario retweeted
“You did in 14 hours what would have taken 7 figures and a year for us to do” - a CTO at a Palantir agentcamp
Rigario retweeted
Tip of the day:
If you need help with configs, settings, setup, usage, exploration, tutorials, or development on Hermes Agent, just type /hermes-agent <prompt>
Built in skill that gives your agent all the knowledge of every CLI command, config, resource links, best practices, architecture info, etc
Even shows your agent how to orchestrate other hermes agent instances and profiles!
Give it a try today :)
Rigario retweeted
Hermes ecosystem is on 🔥
The ecosystem map has been updated to reflect the new v0.8.0 + several new repos added (submit an issue if yours is missing)
Introducing the Hermes Ecosystem Map
I was an early user of Hermes Agent from @NousResearch and have been a power user ever since
But as the ecosystem has grown, its been hard to keep up, so I did some research:
> Scraped every GitHub repo related to Hermes
> Filtered out repos that looked unfinished or had 0 stars
> Built an ecosystem map of everything created and organized it all by category
> Published a website where you can see all the projects with star ratings, and if you hover over you get a short description and link to the repo
Then I had Claude run a security check on every repo to exclude anything that looked sus
Link is in the replies, and also open sourced the repo so feel free to submit PRs if you see anything missing
Oh, and the repo has a /research folder that includes a scrape of everything I could find that's been published on Hermes - you can clone that and add it to your personal knowledge base / wiki
Rigario retweeted
We have partnered with @Xiaomi to bring their excellent MiMo V2 Pro model to Hermes Agent via the Nous Portal - completely free to use for the next 2 weeks!
Access now on the latest version of Hermes Agent: 'hermes update'
Rigario retweeted
In reality, Anthropic’s absolutely idiotic decision to paywall openclaw out of most people’s hands will give way to the other providers to optimize for ultra low cost agentic tokens that give you 95-99% of Claude’s capabilities.
Long term, this means Anthropic is not only toast, but can very easily be seen as the most evil AI company, because their pricing behavior blocks most people out of AGI, only reserving it for the elite - which will cause an acceleration in the wealth gap.
Thank God for competition, because Anthropic is about as morally corrupt as they get. They talk a good game, but their actions are the exact opposite.
Misanthropic indeed.
Rigario retweeted
here is the final verdict you been waiting for. NVIDIA vs Alibaba flagships at full precision on 2x H200 NVL with 280GB+ VRAM.
i tested NVIDIA Nemotron Super 120B and Qwen 122B both on the same benchmark prompt through Hermes Agent. NVIDIA loaded with 1M context, Qwen at 262K. Octopus Invaders is a multi-file JavaScript game. Qwen rendered a screen on first try. game was playable on second try.
when i clicked play and could not enter the game i just told it "game is not playable go check test and fix it" and it debugged and actually made the game playable on the second try at 60 tok/s. as you can see on the uncut video its debugging and fixing in realtime.
for the same test Nemotron took 3 iterations and game rendered. it debugged but game was not playable even though the design and aesthetic was good. model had weight but somehow could not complete. i like Nemotron and i will be doing various testings on these models on upcoming tasks too but so far on GPU marketplace and Octopus Invaders Qwen is undisputed.
one thing i found very surprising is that both 120B MoE models could not one shot the game on enterprise H200 hardware at full precision. something that Qwen 27B dense Q4 one shotted on a single RTX 3090.
this tells us a lot about dense architecture and where the future might be headed. at this point i am more impressed by dense than MoE.
i will continue testing these against new models and later swap the inference engine from llama.cpp to vLLM to see how things perform in terms of speed and GPU utilization and quality. will compare with these base numbers and report back.
this is the lineup that has been winning at every tier i've tested. 27B one shotted a full game on a single 3090. 9B built a playable version on 12GB of VRAM.
now i loaded alibaba's 122B flagship unquantized on 2x H200 NVL and gave it the same job as nvidia. build me a GPU marketplace from scratch.
coded 1,054 lines in a single file from one prompt on hermes agent. one shot. served it successfully to localhost.
Qwen3.5 122B-A10B. pure MoE transformer. 10B active params out of 122B total. 61.5 tok/s at BF16 unquantized across 287GB of VRAM.
full width layouts. particle effects. polished aesthetic. every section with detail. same prompt nvidia got, different result. this model doesn't just build. it finishes.
nvidia coded 769 lines. qwen coded 1,054. same hardware, same prompt, same harness. tonight they both face the real test. octopus invaders. multi-file autonomous build with hermes agent. a game i open sourced so anyone can test their model against the same challenge.
watch the video to the end for the full UI walkthrough.
