Daniel Blasingame retweeted
We had 500+ bankers grade model output on real analyst work. Zero outputs were client-ready. 70% said they'd rather start from scratch than fix the Excel file.
Daniel Blasingame retweeted
In drug discovery and disease research, scientists visually interpret hundreds of experimental results to guide critical decisions.
VIALS evaluates how accurately AI can interpret these same types of scientific images (assays), spanning the types encountered in biotech/pharma workflows.
Current models struggle, lacking scientific knowledge and visual perception, particularly when applying scientific reasoning to images. In contrast, most scientists with the appropriate PhD easily do these tasks in minutes. Until AI can do the same, it usefulness in industry life sciences workflows will remain limited.
We’re launching VIALS-AA, a leaderboard for Handshake AI’s VIALS benchmark for visual reasoning in life sciences and biotech research
Visual artifacts are central to how scientists analyze experiments and make research decisions, but interpreting them correctly remains a challenge for frontier models. VIALS’ 161 tasks span eight domains and need models to reason over artifacts including gel blots, flow cytometry plots, plasmid maps, and phylogenies. Each task requires models to interpret an artifact and draw a specific scientific conclusion.
We evaluate models by running each task three times in Stirrup, our open source agent harness, with code execution and image inspection tools. GPT-6 Luna (medium) grades each answer against the reference answer, and we report the share of tasks judged correct in all three repeats (pass^3).
Key takeaways:
➤ The top three models are separated by 3 points: Gemini 4 Carbon (high) from @GoogleDeepMind leads narrowly with 64%, followed by GPT-6 Astra (max) from @OpenAI at 63% and Claude Opus 5.5 (max with fallback) from @AnthropicAI at 61%
➤ Astra and Sol achieve similar scores to Opus and Sonnet at a much lower Cost per Task: Claude Opus 5.5 (max with fallback) averages $1.64 per task, ~29× as much as Sol (~$0.056). This price difference is significantly driven by turn use - GPT-6 Astra (max) and GPT-6.1 Sol (max) average ~8 turns per task, compared with ~27 for Claude Opus 5.5 (max with fallback) and ~25 for Claude Sonnet 5.5 (max with fallback)
➤ Open weights models trail the highest score by ~19 points: Kimi K3.1 Pro leads open weights models on VIALS-AA at 45%, narrowly ahead of MiMo-V2.6-Pro at 43%
➤ 14 of 161 tasks saw safety refusals on at least one model: Claude Sonnet 5.5 (max with fallback) refused on 6 tasks, the most of any model, and half of the 14 affected tasks saw refusals from multiple models
Thank you to @joinHandshakeAI and the VIALS authors for creating the benchmark and collaborating with us on bringing it to Artificial Analysis!
Daniel Blasingame retweeted
nyc fdes pull up partiful.com/e/oAoel64R8ETWK…
Daniel Blasingame retweeted
Two ATLAS benchmarks are now on @FireworksAI_HQ’s Specialized Intelligence Index, one for finance and one for life sciences, both built with experts from those fields.
ATLAS - Finance puts models through 100 tasks from complex deal work like LBO modeling, structured finance, and M&A. The best model scores 12.3%.
ATLAS - Visual Life Sciences tests whether models understand the images scientists work with in drug development. The best model scores 49.9%.
@nishantXranka on both benchmarks and what we learned building them: joinhandshake.com/research/a…
Daniel Blasingame retweeted
We collaborated with @NVIDIA AI to post-train Nemotron 3.5 Super VL on over 2,000 data points from experts on top enterprise R&D teams, doubling its accuracy on life sciences artifact interpretation.
@joinHandshake is building accurate, efficient models to let researchers accelerate their work and focus on the scientific decisions that drive discovery.
Read the research report: joinhandshake.com/research/a…
Daniel Blasingame retweeted
everyone tells new grads "learn AI." nobody tells them how.
so we built AI Skills Studio with @clay, @figma, @GammaApp, @Google, @Lovable, @NotionHQ, @OpenAI, @Replit, @Salesforce, and @vercel.
free hands on projects, on the same app where 1M+ employers are hiring.
Know you need AI skills but don’t know where to start?
Meet Handshake’s AI Skills Studio. Build free projects with tools employers use, show what you can do and get hired - all in one place.
Built with @clay, @figma, @GammaApp, @Google, @Lovable, @NotionHQ, @OpenAI, @Replit, @salesforce, and @vercel
Ready to build? 🚀 joinhandshake.com/learn?utm_…
Daniel Blasingame retweeted
AI promises to cure disease, but can frontier models interpret the images pharma scientists handle every day?
Best score so far: 26.5% - excited to see this gap to close!
ATLAS Visual Life Sciences (VIALS), a benchmark testing whether AI can interpret the images life scientists make decisions from.
Daniel Blasingame retweeted
Kudos to @anishathalye and @jomulr for co-chairing the RL agentic benchmarks workshop track for the inaugural ACM CAIS conference this week.
We presented two separate Handshake AI Research papers in: (1) AI agentic systems - first evaluation of grader frameworks, and (2) AI benchmarks - first investment banking benchmark. Their posters had big crowds all afternoon. Great job!
Daniel Blasingame retweeted
Replying to @StartupArchive_
Musk is describing the Physics of Value.
"Make more than you take" is the definition of Negative Entropy.
To build is to organize chaos into order. That requires energy and fights the natural drift of the universe.
Extraction (Taking) is easy—it rides the entropy gradient down.
Creation (Making) is hard—it climbs the gradient up.
Money isn't the goal. It's just the thermal exhaust of a high-efficiency engine.
Daniel Blasingame retweeted
Palantir CEO Alex Karp receives the Hudson Institute's Herman Kahn Award.
Intro from @ssankar and full speech from Dr. Karp below.
Daniel Blasingame retweeted
Palantir honors those who swore the oath and kept the faith. Happy Veterans Day.
Daniel Blasingame retweeted
IN NEWS: Palantir just closed its first-ever $1B deal outside the US.
@louismosley (EVP of Palantir, UK) told us: “We are the operating system for the modern battlefield… the significance of today is the UK, America’s closest ally, making the same move.”
Alongside the deal, Palantir announced a $2B investment into the UK and 350 new jobs.
Palantir is deeply committed to the UK and proud to deepen our partnership with the Ministry of Defence and serve the country’s brave armed forces.
Today’s agreement will pave the way for Palantir’s advanced AI-enabled defence technology to be deployed on an unprecedented scale with up to $1bn investment by the UK in our military software over the next five years.
We are also proud to announce plans to invest up to $2bn in our UK business over the same period and make London the European Headquarters of our Defence business.
Daniel Blasingame retweeted
“You can make your organization more effective without compromising on security and privacy through the use of technologies like Markings, Cipher, Checkpoints, and granular access controls"
See how Palantir AIP enables transformation while protecting privacy and civil liberties.
Chad & Arnav | Privacy & Security with Palantir AIP
Daniel Blasingame retweeted
$PLTR
PALANTIR INTRODUCES MMDP:
Multi Modal Data Plane.
Forward Deployed Architect @chadwahl joined me to explain the future of interoperability and what it means for businesses that want to get the value of Palantir but also don't want to leave their existing tech stack.
This could be an absolute game changer for the adoption of Palantir's software because the core focus of MMDP is on providing outcomes, regardless of whatever model or compute you use.
Most SaaS companies want to lock you into their tech stack in order to continue charging you, but if Palantir can meet you where you are and provide outcomes as the basis for why they deserve to get paid, we could see an incredible adoption rate as more businesses don't have to leave their existing stack just to access Palantir.
More details:
Daniel Blasingame retweeted
On the @USArmy’s 250th birthday, Palantir honors the indomitable spirit of the American Soldier.
From Valley Forge to Shiloh, from Normandy to today’s far-flung battlefields.
Count the brave. Count the true.



























