Entusiasta de la Tecnología. Fundador @AIStudioLab. 🚀🐕🦺
Madrid
Joined January 2016
- Tweets4.4K
- Following410
- Followers35
- Likes12.9K
Gabe retweeted
US open-weights models made gains in both intelligence and parameter efficiency last week:
➤ Thursday morning: NVIDIA released Nemotron 3.5 Bolt and Nemotron 3.5 Super VL. Super VL scored 26.5 on the Intelligence Index with 120B parameters, pushing Inkling Small off the Pareto frontier
➤ Thursday midday: Meta's Muse Spark Open scored 38.9, more than 12 points ahead of Super VL and the highest intelligence yet for a US open weights model
➤ Friday morning: Beam 1.0, Reflection's first model, scored 30.3 with 501B parameters, joining the frontier between Super VL and Muse Spark Open
See our full comparison of open weight models on Artificial Analysis artificialanalysis.ai/models…
Gabe retweeted
4. Aprender rápido
Las herramientas cambian cada pocos meses.
Quien se adapta en semanas y no en años va a tener ventaja siempre.
Gabe retweeted
This is a beautiful blog post on how neural networks are able to remember things.
Looking at linear layers, various forms of MLPs and self-attention from a memory perspective is surprisingly illuminating. Highly recommended reading!
Gabe retweeted
Using Claude? -> Anthropic releases Claude Opus 5.5, Sonnet 5.5 and Haiku 5.5
"On our automated behavioral audit, the most comprehensive alignment test we run, Opus 5.5 is the strongest-performing model we’ve tested to date. It also comes with the safeguards we’ve developed for our most capable models."
"Opus 5.5 is a major step up from Opus 5. It’s the new leading model, and early testers saw large jumps in performance on their most complex work. One tester completed a 680,000-line code migration in less than a day—work that would have taken an engineering team weeks."
anthropic.com/claude-opus-5-…
Gabe retweeted
This is really a wild way to get adoption on your models.
Really great for the community, and the timing is hilarious given the govt just banned deepseek in federal agencies.
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable.
🔗 github.com/deepseek-ai/DeepI…
1/4
Gabe retweeted
Exa has released Exa 3.0, scoring 80 on the Artificial Analysis Search Index in Dynamic Highlights mode, tying with Perplexity Search for the #1 spot
@ExaAILabs has released a new Dynamic Highlights mode to its Search API. Given a search query and its results, Dynamic Highlights determines how much text to return from each page within a shared budget
We benchmarked two Exa search types with Dynamic Highlights ahead of release: an upgraded auto mode, which balances search quality and speed, and a new lite mode, optimized for lower cost and latency. Both were evaluated using our Stirrup agent harness and GPT-5.6 Luna (medium) as the agent model
Key benchmarking results for Exa Search with Dynamic Highlights:
➤ Ties for #1 on the Artificial Analysis Search Index: Exa Search (auto, Dynamic Highlights) scores 80, level with Perplexity Search (medium) in the #1 spot, while Exa Search (lite, Dynamic Highlights) scores 79. On DeepSearchQA, auto (85.6 F1) and lite (84.9 F1) rank 1st and 2nd respectively, 4 to 5 points ahead of Perplexity Search (medium, 80.8)
➤ Reduces search calls and token usage: Exa Search (auto, Dynamic Highlights) uses 7.6 searches per task, the fewest of any Search API on the leaderboard, down from 9.4 with standard highlights. Tokens per task fall 47%, from 299k to 157k
➤ Reduces cost per task by 21%: Exa Search (auto, Dynamic Highlights) costs $0.066 per task, compared to $0.083 with standard highlights. Lite costs $0.044 per task, with both sitting on the cost vs. capability Pareto frontier. Exa Search is priced at $7 per 1k searches for auto and $4 per 1k searches for lite
Gabe retweeted
We keep adding skills to coding agents. But how often do agents actually use them?
A paper called Cortex studies how to improve coding-agent instructions through measurement.
The researchers tested rules and skills using replay-based evaluations.
One finding caught my attention.
Path-scoped rules produced most of the improvements in following repository conventions.
Skills that appeared on every turn were rarely opened unless they matched the files the agent was working on.
This has a practical implication.
A large collection of skills doesn't guarantee that an agent will use them.
The harness needs to decide which instructions matter for the current task.
For example:
Editing database migrations?
Load migration rules.
Changing authentication?
Load security and permission rules.
Working on frontend components?
Load the project's UI conventions.
I would rather measure whether a skill improves verified task completion than count how many skills an agent has installed.
I'd also remove rules that don't improve results.
Every instruction consumes context, and unnecessary instructions can create conflicts.
Cortex suggests treating agent rules as something we can test, improve, and prune.
Gabe retweeted
The Marine Forces Reserves is seeking a Palantir Maven Field Service Representative.
GRAB A KLEENEX - PTFB
The Marine Forces Reserves is looking for a Forward Deployed Engineer & Trainer to provide specialized information technology systems architecture, data engineering, artificial intelligence enablement, and operational application delivery services in support of the Marine Forces Reserve.
This includes, but is not limited to, Maven Smart System and War Data Platform build and sustainment, governed ontology engineering, Databricks-native pipeline development, AIP Logic and agentic workflow integration, rapid operational prototyping, and organic workforce training for reserve Marines and civilian personnel.
To accelerate operational decision-making, bridge data silos, and field modern data/AI capabilities across the Total Force, Marine Forces Reserves requires one cleared forward-deployed engineer embedded directly with MFR staff.
This individual represents the advanced engineering equivalent of a Palantir Maven Field Service Representative, possessing the technical depth not only to operate and support fielded software platforms, but to actively build, integrate, and deploy data pipelines, ontologies, and operational applications on live operational networks.
Gabe retweeted
Happy Monday. On today's show:
- @bingfish (Kleiner Perkins)
- @dakshgup (Greptile)
- @jgreze (Town)
- @quintendf (Tolan)
- Katie Drummond (WIRED)
See you on the stream.
50 Gigawatts To The Face, 17% GDP Growth Please, Road to COD 4, Bing Gordon Joins xitter.cf/i/broadcasts/1NGaroZpr…
Gabe retweeted
ALERT🚨: NVIDIA has been working overtime, nights, and weekends after being shocked 🤯 by last week’s TPUv7 FP8 submission showing better performance per dollar.
Unfortunately for NVIDIA, its first GB300 FP4 counter-submission is still 1.64x worse in performance per dollar than TPUv7 for external customers across most of the end-to-end performance curve.
The x-axis of the graph below uses end-to-end latency, so a higher curve is better. The TPUv7 curve has higher perf per $ across most of the E2EL Pareto frontier.
This initial submission looks like NVIDIA’s first attempt to close the gap, but the gap is still massive. We expect more counter-submissions from NVIDIA as they continue trying to narrow the difference.
Gabe retweeted
Google has been named a Leader in the inaugural @IDC MarketScape: Worldwide Agentic AI Platforms 2026 Vendor Assessment.
Read the excerpt to learn why we are a Leader → goo.gle/3VRCVoV
Gabe retweeted
Exciting news: Muse Wonder Full (Max) by @AIatMeta has just debuted in the Code Arena: WebDev at #6 overall with 1726 pts, putting Meta in the top 3 labs!
This is +75 pts and 11 places above Muse Wonder (Max) at #17 (1651 pts), and +67 pts and 8 places above Muse Spark 1.3 (Max) at #14 (1659 pts).
Muse Wonder Full (#6) also landed +16 pts above Claude Sonnet 5.5 (#17), and +48 pts above Gemini 4 Argon (#10). With this score, Muse took the #3 lab spot (previously held by Google).
Across categories, Muse Wonder Full (Max) landed:
- #4 in Data & Analytics and in Consumer Product
- #5 in Brand & Marketing
- #6 in Content Creation Tools
- #7 in Gaming and in Simulations
- #8 in Reference-Based Design
Congrats to the @AIatMeta team and @alexandr_wang on this release!
Muse Wonder (Max) by @AIatMeta has entered Code Arena: WebDev at #15 with 1651 pts and Text Arena at #2 with 1513 pts!
In Code Arena: WebDev, Muse Wonder (Max) landed at #15 with 1651 pts. This result is slightly lower thanMuse Spark 1.3 (Max) at #12 with 1659 pts, within 4 pts of Kimi K3 (Max) at #14 with 1655 pts, and 11 pts above Grok 4.7 (xHigh) at #16 with 1640 pts.
This release is a significant improvement over past Muse models in Text Arena. Compared with Muse Spark 1.3 (Max) at #11 with 1494 pts, Muse Wonder (Max) gained +19 pts and 9 places to #2. It's 12 pts behind the #1-ranked Gemini 4 Argon (High) with 1525 pts.
Congrats to the @AIatMeta team on this release!
Gabe retweeted
Introducing Clairic, from Cloudflare. A System One extraction modeI, capable of extracting, computing, estimating, and forecasting data from unstructured inputs. cfl.re/4zIgfpH
Gabe retweeted
But hold on, at least there's an E-E-A-T policy... Oof geroundtable.com/editorial-p…
They say "imitation is the sincerest form of flattery" - this new site just copied my design, my features and some of my content from seroundtable.com/
Gabe retweeted
WAIT.. 1¢ PER PROMPT TO FIND OUT WHY AI RECOMMENDS YOUR COMPETITORS?!
🚨 @treg_ai’s new open-source AEO toolkit shows you the pages behind those answers and the people who wrote them:
→ see which sources AI cites
→ find the people behind them
→ draft outreach to get your product considered
.. and more!
run it from Claude Code. or Codex.
Self-hostable. No subs!
Best part?
Open source. Full code on GitHub. Repo in 🧵↓
Introducing OpenSource AEO Toolkit
- Find every source AI cites, and the author behind it
- Get your product into the sources AI trusts
- 1¢ per prompt, no subscription
- Access 4000+ data & services
Try it at treg.to/aeo
Partner with @cloroai
Repo link below 👇
Gabe retweeted
Can Agents improve their own harness on unseen tasks? The @GoogleResearch team released Harness², a framework that allows improve an harness around a frozen model.
How it works:
1️⃣ Agent runs the base harness on each task to record a baseline trajectory
2️⃣ Editor agent inspects the trajectory, drafts 2 candidate edits (SI, Skills, hooks,….), and runs both.
3️⃣ Composer agent diffs the new trajectory against the baseline, merges edits that fixed failures (within a cap), and runs the final harness.
4️⃣ Reflector agent logs verified lessons into file and saves the winning components to a library for future tasks to pull on demand.
Results:
- 💡 Gemini 3.5 Flash +8.4 on WorkBuddy-Bench and +7.5 on JobBench.
- 🛠️ Skills and scripts drive ~50% of edits; dropping skills costs 5.3 pts.
- 📚 Gemini 3.5 Flash improves on all 3 benchmarks when editing its own harness.
Paper: arxiv.org/abs/2610.14224
Code: github.com/google-research/h…
Gabe retweeted
es que le puedes pedir hasta que te introduzca animaciones en los menús y te lo one-shotea
básicamente puedes hacerte tu SO a medida
Gabe retweeted
Si estuviera en educación, revisaría este dataset y crearía cursos personalizados con IA para cada niño de primaria. Vivimos en una era en la que esto ya es posible en hacerlo en cualquier idioma. Educadores 👀👇
In July we open-sourced a dynamic graph of everything a child learns in primary school. 1,590 concepts, 3,221 dependencies.
Now we’re turning every concept into interactive experiments kids can play and learn from.
First one, age 7: heat and water.
Everything that used to be a boring textbook can now be an engaging experience, potentially tailored to every child.
Kids education just got a massive upgrade.