Developer, interested in many different tech fields. Working at Ubisoft Paris.
- Tweets 17,929
- Following 214
- Followers 210
- Likes 23,428
Manu_TechAndGames retweeted
I published a new blog post explaining the steps I used to reverse engineer the original "Comanche: Maximum Overkill" terrain/map files.
🔗pikuma.com/blog/comanche-map…
It seems like, lately, anything related to decompilation, reverse engineering, & other retro game stuff has AI written all over it. Even though I get it, it all feels as good as drinking water when I'm not thirsty.
Not this time! I've literally decided to hide myself off-grid using only a 486 with no internet for this one. 10/10 would recommend! 😂
8
22
207
4,305
Got this email and was genuinely surprised by this point of view. With all respect but do people like this plan to live this world.
50
4
327
21,174
This happens at scale with video game studios and you can tell by the quality of video games nowadays how much they are controlled by their own audience
1
3
653
Manu_TechAndGames retweeted
When I stilled worked on biosecurity, I first learned about Debug, a Google project to reduce mosquito populations.
Results from their Singapore pilot are now out.
They crushed mosquito populations, reducing dengue infections by 70% in the process:
There is no good reason for mosquitoes to exist in wealthy countries today.
New technologies can get rid of the mosquito species that cause diseases like dengue, yellow fever, and malaria; and others can prevent them from carrying those diseases.
worksinprogress.co/issue/mos…
So why do mosquitoes still exist?
In a new article for Works in Progress, Maya Rosen explains how state incapacity and risk aversion led to over a decade of delays and billions of unnecessary bites in the United States.
Oxitec, for example, engineered an Aedes aegypti mosquito in such a way that its female offspring would die before adulthood. Through sustained releases, they showed they could reduce local mosquito populations by 80% or more, in field trials.
They sought approval from the USDA in 2010, which then passed it to the FDA, which passed it to the EPA. In 2020, the EPA allowed them to conduct limited field trials in the Florida Keys. Two years ago, their experimental permit expired, and they still need another review before it can be used routinely. Their proposed field trial in California was withdrawn after public opposition.
Another company, MosquitoMate, developed a technique involving Wolbachia, which also suppresses local mosquitoes. They then spent over fifteen years working through the system before finally receiving commercial registration nationwide in 2024.
It seems that, sometimes, the barrier to getting rid of diseases isn’t technology at all. It’s policy and political will.
New products shouldn’t get stuck in regulatory review for so long, especially if the delays are about basic questions like which agency has oversight over the product, rather than testing for safety.
Though some of those uncertainties were being addressed, the broader changes haven’t materialized. After Biden’s 2022 executive order, the agencies planned to update how their coordinated framework was implemented. But Trump rescinded the order last year.
Meanwhile, the EPA unit that reviews new mosquito control products is still a small office backlogged in reviewing a wide range of products.
Mosquitoes are a policy choice. We can get rid of them. Will we choose to?
worksinprogress.co/issue/mos…
73
542
5,086
440,351
Decision models are the new moat...
We turned Qwen3.8-27B into a multimodal decision model.
It beat Pokémon FireRed’s elite four and champion with sub-100 ms decisions from live game state.
With SGLang’s native /v1/decisions, you can now turn LLMs and VLMs into classification and scoring models.
We also added /v1/systemone so Jev-like open models can work with the TypeSafe SDK.
1
17
Manu_TechAndGames retweeted
So I tested GPT-6.1 Sol on real work: 2 repos, 105 planted bugs. Find and fix what you can.
Unlike GPT-6 Sol, which was just a nerfed GPT-5.6 Terra, this one is real. The results:
- GPT-6 Astra (max): 45 for $33
- GPT-6.1 Sol (max): 44 for $6.56
- GPT-5.6 Sol (max): 43.5 for $95.35
- Opus 5.5 (max): 41.7 for $58.53
- GPT-6 Sol (max): 29.3 for $9.33
With a model like this, the 50% cut to the $200 plan doesn't matter.
n=1. More runs and results at more effort levels dropping in this thread over the next few hours 🧵
211
195
2,889
299,638
Manu_TechAndGames retweeted
We rolled out a new Kimi K3 inference optimization autoresearch loop that ran on 120 B300s and found a 40% efficiency gain in < 6 six hours
Compute cost = 120*6*$5.80/hr =$4,176
Eng + token cost = ~$5k
Total cost = $9,176
Will pay itself off in < 12 hours at current scale
29
27
893
51,018
we found opus 5.5 to be 2X as efficient as gpt-6-sol in our internal cfo.ai evals so i was not expecting 6.1-sol to improve on that much
but to see it 3X the efficiency of opus 5.5 and 2.5X against even sonnet 5.5 is actually 🤯
this is an insane point release
112
198
3,450
224,546
Manu_TechAndGames retweeted
I open-sourced Dioramas, a framework for cinematic, interactive 3D websites.
It's 100% free. No paid tier, no signup. It comes with 20 full example sites, each built around one interaction you can touch: a museum you explore by lantern, a mech you power up, a geode you crack open.
It's powered by fal: every 3D model was generated with Meshy 7.1 from Nano Banana 2 images. Every prompt, model and line of code is in the repo.
Repo: github.com/blendi-remade/dio…
13
77
632
23,814
Manu_TechAndGames retweeted
ALERT 🚨🚨OpenAI's latest GPT6.1 Sol Ultrafast is NOT running on Cerebras but is instead running at a low batch size on NVIDIA GPUs.
What does this say about Cerebras? Will Cerebras be serving GPT6.1 Sol Ultrafast in the future?
117
79
1,668
753,908
Manu_TechAndGames retweeted
New blog post on speeding up the separating axis test (SAT) using inscribed spheres: box2d.org/posts/2026/09/insc…
3
20
142
5,047
@AndroidBlogger tu sais ce que c’est cet orchestrateur et cette UI ? Je veux
Claude Code tip: once Opus 5.5 is your main model, stop leaving Fable 5.1 sitting idle
put it on call with /advisor
run /advisor fable
Opus 5.5 keeps writing the code
Fable 5.1 reads the full session, every tool call included, and only speaks up at three points:
→ before a plan: is this the right approach?
→ when the same error comes back: am I digging in the wrong place?
→ before "done": what did I miss?
Fable 5.1 reviews. Opus 5.5 ships
Jev engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to Jev in under half a second, and the big model only sees the ones that split
- the full tree
> Opus 5.5 on high runs the main session
> explorer reads the code
> worker edits and runs tests
> researcher pulls the docs
> all three on medium
> Fable 5.1 on call as the advisor
paste the tree and this prompt into Claude Code ↓
"Rebuild my Claude Code setup around this tree:
1. Check ~/.claude/agents and .claude/agents for subagents that already fit explorer, worker and researcher.
> Draft new ones only for missing roles
> Give each model: opus, effort: medium
> Skip any that pin a different model and list them
2. Set the main session to high via effortLevel in ~/.claude/settings.json, and set advisorModel to fable
3. Find anything that keeps the advisor off (CLAUDE_CODE_DISABLE_ADVISOR_TOOL, DISABLE_TELEMETRY, any variable that stops feature-flag fetching) plus CLAUDE_CODE_EFFORT_LEVEL, which overrides subagent effort. Report them, change nothing
4. Add one rule to ~/.claude/CLAUDE.md: consult the advisor before a large plan, when an error repeats, and before calling a long task done
Show me every change as a diff first. No edits until I say go."
↳ code.claude.com/docs/en/advi…
1
10
Manu_TechAndGames retweeted
you're not crazy
between anthropic and openai, a new model used to come out every ~10 weeks
now it's every ~11 days
138
422
6,356
283,787
Manu_TechAndGames retweeted
Artificial Analysis numbers are in! GPT-6.1 Sol is performing at around Opus 5.5 Medium levels for under a third the price.
Not bad at all, but still more hyped for 6.1 Astra to catch them up 👀
83
65
1,749
151,522
Manu_TechAndGames retweeted
One interesting @OpenAI detail: researchers let models continuously optimize the computer-use harness itself, then feed the successful improvements back into training.
OpenAI says this has already cut latency by more than 2×.
AI helping improve the systems used to train AI is getting very concrete.
3
2
5
2,642
Manu_TechAndGames retweeted
OpenAI just announced Ultrafast a model that can run up to 300tok/s ! 8x faster .
I’ve been testing the new Ultrafast for some time !
I had two agents playing Tekken on PS1 just by looking at images. They made one decision every 0.95 seconds.
That enables for the first time near-real-time, high-intelligence interactions.
This unlocks a bunch of new use cases!! 🚀
8
5
90
9,908
Manu_TechAndGames retweeted
That’s a great price reduction but what I would also love see is how I can delegate the choice of the right-for-task model to an automated routing.
1
2
8
2,649
Manu_TechAndGames retweeted
🚨 BREAKING: Super excited to announce we're deploying the world's fastest inference with @Cerebras.
Talk to any developer and they're excited to build with 20x faster AI... the problem is there's almost no compute available.
We're here to solve that.
Using GPUs for prefill - it's now more affordable than ever too.
Thank you to the whole Cerebras team and excited to grow this partnership.
First tokens live Q127 🚀
66
128
987
676,726
