Ryan Marten ✓
@ryan_marten
Building @harborframework and @terminalbench
3 270Followers
2 253Following
686Posts total
89.8KViews on collected posts
Derniers posts
Recherche de nouveaux posts…
@modal_labs @AnthropicAI @OpenAI @Google @scale_AI @SnorkelAI @turingcom @gNucleusAI @joinHandshake @ellamindAI @giansegato @boolean_ai Thank you to @modal for providing the sandboxes for developing Frontier-Bench! This was particularly useful for unlocking GPU tasks (e.g. agent
1.7K views · 12 likes
· 0 reposts · 1 replies
Open on X →
@modal_labs @AnthropicAI @OpenAI @Google @scale_AI @SnorkelAI @turingcom @gNucleusAI @joinHandshake @ellamindAI @giansegato Thanks as well to @boolean_ai (missed the tag above). It was a pleasure working together!
1.8K views · 12 likes
· 0 reposts · 1 replies
Open on X →
@modal_labs @AnthropicAI @OpenAI @Google @scale_AI @SnorkelAI @turingcom @gNucleusAI @joinHandshake @ellamindAI Another thank you to @giansegato as an advisor to this project - making sure measurement in benchmarks is rigorous as possible!
1.9K views · 11 likes
· 1 reposts · 1 replies
Open on X →
@ryan_marten Curious the score with Grok 4.5 with Grok Build harness.
2.3K views · 34 likes
· 0 reposts · 0 replies
Open on X →
@modal_labs @AnthropicAI @OpenAI @Google @scale_AI @SnorkelAI @turingcom @gNucleusAI @joinHandshake Additional thank you to another partner of ours who was instrumental for this release. @ellamindAI led the expansion of tasks, via creation and reviews, into the business operation
2K views · 12 likes
· 0 reposts · 2 replies
Open on X →
Frontier-Bench was made possible by our sponsors.
Thank you to our compute sponsors @modal_labs, @AnthropicAI, @OpenAI, @Google and our data partners @scale_AI, @SnorkelAI Open Benchmarks, @turingcom, @gNucleusAI, Boolean AI, and @joinHandshake.
Frontier-Bench is hosted by
3.3K views · 42 likes
· 5 reposts · 2 replies
Open on X →
@frontierbench is built by the @terminalbench team and @harborframework community
Tasks are the beating heart of any worthy benchmark. Thank you to all the task creators and reviewers for all their hard work making this first release. We are lucky enough to have too many
2.9K views · 33 likes
· 0 reposts · 1 replies
Open on X →
Frontier-Bench is a community effort from 100+ task contributors and reviewers. It is the hardest and highest quality benchmark we could make together in the open, but it is just the start.
Let's find issues, fix them, and add new tasks together. Let's make this benchmark
3.1K views · 35 likes
· 0 reposts · 1 replies
Open on X →
Special thanks to our senior reviewers @neversupervised, @bla1990so, @TommasoCerruti, @StevenDillmann, @ryne_wang, @dwahdany, @AllenHa13152844, @krauth, advisors @Mike_A_Merrill and Nicholas Carlini and co-leads @alexgshaw, @andykonwinski, @lschmidt3.
2.6K views · 27 likes
· 1 reposts · 1 replies
Open on X →
The Frontier-Bench team is already planning upcoming releases. In our next major release, we plan to add new tasks that measure new capabilities, add greater task diversity, and tune timeouts.
Tasks are versioned so corresponding trials can be re-used, re-graded, or re-run to h
3.2K views · 37 likes
· 1 reposts · 1 replies
Open on X →
Agents approach problem solving on Frontier-Bench differently, even at similar pass rates.
Fable 5 (33.8%) and Opus 4.8 (21.1%) use more tokens and take fewer actions.
GPT-5.6 Sol (34.4%) and Terra (20.8%) use less tokens and take more actions.
Overall, GPT-5.6 Sol is ~40% htt
4K views · 43 likes
· 0 reposts · 3 replies
Open on X →
Frontier-Bench discriminates better between frontier and sub-frontier models than Terminal-Bench 2.1. For example, Fable 5 in Claude Code and Opus 4.8 in Claude Code are separated by only 4.9% on Terminal-Bench v2.1, but 12.7% on Frontier-Bench v0.1. https://t.co/gsYh89JRpt
4.7K views · 51 likes
· 0 reposts · 2 replies
Open on X →
Tasks in Frontier-Bench represent domains like software, ML, science, operations, security, hardware, and media.
Every task separates the agent container and verifier container. Artifacts from the agent container are downloaded at the end of the trial, logged, and uploaded to ht
5.2K views · 57 likes
· 0 reposts · 1 replies
Open on X →
Frontier-Bench is a continuous benchmark. Today we are releasing v0.1 with 74 difficult, diverse, and high quality tasks. We will be adding new tasks and improving existing ones in regular releases.
Benchmarks have historically depreciated over time. We’re rolling out a batch of
9.9K views · 65 likes
· 0 reposts · 5 replies
Open on X →
Growth & engagement
How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.
Views per post
9.9K23 Jul
5.2K
4.7K
4K
3.2K
2.6K
3.1K
2.9K
3.3K
2K
2.3K
1.9K
1.8K
1.7K
Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.
Engagement rate per post
0.72%23 Jul
1.12%
1.15%
1.15%
1.21%
1.10%
1.17%
1.18%
1.51%
0.74%
1.46%
0.67%
0.71%
0.76%
Reactions — likes, reposts, replies and quotes — divided by views.
What the audience does
Likes85.3%614 in total
Reposts3.1%22 in total
Replies4.0%29 in total
Quotes2.5%18 in total
Bookmarks5.1%37 in total
Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.
The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.
Comptes similaires
Shyaka Lixy🇺🇸 ✓
@ShyakaRay
6.5K followers
· 51.7K posts
Foreign Press Association in London
@FPALondon
6.5K followers
· 3.3K posts
Bukhosini ka Ndaba
@Shaka66622964
6.5K followers
· 9.9K posts
Michael Farris
@Michael_Frdt
6.5K followers
· 132.7K posts
فرقة أنصار الله الإنشادية
@AnsarAllahB
6.5K followers
· 330 posts
Suhan Samyeli
@SamyeliHilmi
6.5K followers
· 2.4K posts
Ben Tavener ✓
@BBCBenTavener
6.5K followers
· 15.3K posts
Alice Han ✓
@alicekpop_ai
6.5K followers
· 2K posts