Benchtalks #2: @jyangballin
, creator of SWE-bench, fresh off ProgramBench (every frontier model: 0% at launch). Out soon with @vincentsunnchen
.
@snorkelai
-
Benchtalks #2: SWE-bench creator jyangballin discusses ProgramBench and upcoming project
By
–
-
AI capabilities outpace measurement, says Snorkel AI CEO
By
–
“Our ability to do stuff with AI has significantly outpaced our ability to measure capabilities,” says @ajratner, CEO of Snorkel AI.
— Snorkel AI (@SnorkelAI) 31 mai 2026
Alex recently joined the @chain_ofthought podcast with @ConorBronsdon to talk all things frontier AI.
Watch the full conversation covering the… pic.twitter.com/cQZkRoY16i“Our ability to do stuff with AI has significantly outpaced our ability to measure capabilities,” says @ajratner
, CEO of Snorkel AI. Alex recently joined the @chain_ofthought podcast with @ConorBronsdon to talk all things frontier AI. Watch the full conversation covering the -
JudgmentBench: First Public Benchmark for AI Quality Assessment Released
By
–
Congratulations to @StanfordLaw liftlab on the release of JudgmentBench, the first publicly available benchmark in a high-judgment domain where both methods for assessing quality are solicited over the same tasks. Snorkel was proud to contribute as research and data partners on
-
Frontier AI Models Struggle with Long-Horizon Legal Tasks, Study Finds
By
–
Initial LAB results from Harvey put a number on something we see across specialized AI work: under rigorous all-pass standards, frontier models complete fewer than 10% of long-horizon legal tasks, and no single model leads across practice areas. General capability isn't
-
Supporting Terminal-Bench Science for Scientific AI Benchmarking
By
–
We're proud to support @StevenDillmann on Terminal-Bench Science, an open benchmark pushing the frontier of scientific AI through Open Benchmark Grants. But it needs YOU – the research community – to shape the next wave of computational science workflows. → Contribute here:
-
Methodology for steering AI model behavior
By
–
“If you want your model to do something different, you don’t go and perform brain surgery in the neurons. You don’t reinvent the architecture of the algorithm. You don’t say ‘hey, I want my agent to be better at cobol in hospital systems, let me go and invent a new… pic.twitter.com/nhXZxntqaS
— Snorkel AI (@SnorkelAI) 16 mai 2026“If you want your model to do something different, you don’t go and perform brain surgery in the neurons. You don’t reinvent the architecture of the algorithm. You don’t say ‘hey, I want my agent to be better at cobol in hospital systems, let me go and invent a new
-
Frontier AI Models Push Limits in Geospatial Applications
By
–
For AI in geospatial, "there are a lot of different research areas that have to come together to build this one system. We're literally running into the limits of frontier models." — @intellgenc of @usachp in conversation with @SniffenOutAI.
— Snorkel AI (@SnorkelAI) 15 mai 2026
Full context of what he's cooking:… pic.twitter.com/uwVaCgUAe7For AI in geospatial, "there are a lot of different research areas that have to come together to build this one system. We're literally running into the limits of frontier models." — @intellgenc of @usachp in conversation with @SniffenOutAI
. Full context of what he's cooking: -
Building an AI-Native Geospatial and World Simulation System
By
–
"We're working with Google and @SnorkelAI to build a geospatial, deep-research, AI-native system. And along with that…we're building a world simulation system to simulate impacts for large infrastructure projects." — Rezaur Rahman, CIO / CISO / CAIO, @usachp
— Snorkel AI (@SnorkelAI) 14 mai 2026
Full conversation… pic.twitter.com/4QhdGYXsmx"We're working with Google and @SnorkelAI to build a geospatial, deep-research, AI-native system. And along with that…we're building a world simulation system to simulate impacts for large infrastructure projects." — Rezaur Rahman, CIO / CISO / CAIO, @usachp Full conversation
-
Workshop on Reliable Financial Agents at AICouncilConf
By
–
We're live from @AICouncilConf A huge thank you to everyone who joined Charles Dickens’ workshop this morning: Towards Reliable Financial Agents: How a 4B Model Outsmarted a 235B Giant. You can find our team in the exhibitor hall throughout the rest of the conference.
-
AI Council Conference Participation and Financial Agent Workshop Announcement
By
–
We’ll be at @AICouncilConf in San Francisco this week (May 12–14). Come say hi if you’re attending Join us tomorrow from 10:15–11 AM for a workshop with Charles Dickens: Towards Reliable Financial Agents: How a 4B Model Outsmarted a 235B Giant. See you there: