Applied Compute @appliedcompute

@sigalitperelson so back

@_gracehua Thanks for coming!

“A modelless company is sitting on shifting sand.” Our CEO @ypatil125 sat down with @mariogabriele to talk about why owning your model is the difference between building on bedrock vs on someone else’s roadmap. It’s the core of what we do at Applied Compute. We train better, faster, cheaper custom models on your data, serve them in production, and continuously improve them as your definition of “good” evolves. Your model becomes your moat instead of a dependency that can change underneath you. Full conversation below!

Mario Gabriele @mariogabriele

a day ago

"A modelless company is sitting on shifting sand." Yash Patil (@ypatil125) is the founder and CEO of @appliedcompute, a company that trains custom models on company data and serves them in production. His conviction: every organization has its own definition of what good looks

2 8 60 20K 61

Harvey partnered with @appliedcompute to train a legal agent. We optimized each part of the agent stack: - eval loop - agent harness and compaction - post-trained GLM-5.1 using reward signal from our Legal Agent Benchmark (LAB) More in our agent-training deep dive:

Harvey @harvey

2 days ago

x.com/i/article/2069…

2 12 146 93K 223

Harvey partnered with @appliedcompute to train a legal agent. We optimized each part of the agent stack, including the eval loop, agent harness and compaction, and post-trained the underlying GLM-5.1 model using reward signal from Harvey's Legal Agent Benchmark (LAB). Check out more in the agent-training deep dive below. Kudos to @nikogrupen, @ItsJulioPereyra, @rhythmrg, @jacob_dphillips, and @raymondmfeng for leading this effort - more to come, with lots of opportunity to push the frontier with GLM-5.2.

Harvey @harvey

2 days ago

x.com/i/article/2069…

2 12 146 93K 223

It was great collaborating with @nikogrupen, @ItsJulioPereyra, and @gabepereyra on a custom post-trained model for LAB. The rigorous work Harvey is doing to map out and build representative evals that reflect how real legal work gets done will pay massive dividends over time and help them continue to build unique research and model IP.

Applied Compute @appliedcompute

2 days ago

We partnered with @harvey to post-train the state-of-the-art legal agent on their LAB benchmark. It surpasses Opus 4.8 Max and GPT-5.5 xhigh.

1 10 91 24K 28

More evidence that the frontier is attainable with (1) high quality environments and domain expertise, of which @harvey has in abundance; (2) post training infrastructure to execute big runs. This ran on our Blackwell cluster without any issues, thanks to infrastructure that allows us to elastically scale inference compute. Was great to collaborate with @gabepereyra @nikogrupen and team!

Applied Compute @appliedcompute

2 days ago

We partnered with @harvey to post-train the state-of-the-art legal agent on their LAB benchmark. It surpasses Opus 4.8 Max and GPT-5.5 xhigh.

1 10 91 24K 28

If we’ve learned anything this past week it’s that GLM is a strong base for customization. Together with @appliedcompute, we focused on GLM-5.1 and have results that are a great example of what full-stack agent optimization looks like —> post-training + harness + verifier. Real potential to push the Fable frontier now with GLM 5.2. Great working with @ypatil125 @rhythmrg & team!

Harvey @harvey

2 days ago

x.com/i/article/2069…

2 12 146 93K 223

We partnered with @appliedcompute to train a legal agent. We optimized each part of the agent stack: - the evaluation loop - the agent harness - and post-trained the underlying GLM-5.1 model. The result? The agent achieved the highest rubric pass rate on our Legal Agent Benchmark (LAB) of any available model. Much more in our agent-training deep dive with @appliedcompute:

keyboard_arrow_left Previous keyboard_arrow_right Next

Harvey @harvey

2 days ago

x.com/i/article/2069…

2 12 146 93K 223

Read the full report: appliedcompute.com/case-studies/h…

We rebuilt the agent harness to operate well in challenging long context environments. Legal source documents are huge, with the 90th-percentile LAB task carrying nearly 100k tokens and some exceeding 200k. We added compaction so the model summarizes its own transcript and continues past its context limit, letting it work far longer on a task without losing the thread.

We partnered with @harvey to post-train the state-of-the-art legal agent on their LAB benchmark. It surpasses Opus 4.8 Max and GPT-5.5 xhigh.

Model strategy for @harvey: We are working on the first model in our legal foundation model series, inspired by @cursor_ai's Composer. Two goals: 1. Allow us to serve frontier intelligence across our product surface areas at an affordable price and a strong security posture. 2. Create the foundations for law firms to build their own specialized models and own their own intelligence. The model series will focus on complex client matters that span months and take dozens of associates. The agentic system will learn to control legal tech tools, sub agents and ask for help from frontier models or human partners, much like a senior associate. We’ve open sourced benchmarks for evaluating our initial post training work that represents work done by associates and in-house lawyers. We are scaling these significantly using synthetic and human pipelines as well as building private evals for firms. Open sourcing this data has allowed us to quickly validate the feasibility of post training open weight models for legal work. With our research partners we’ve already shown promising results post training open source models to approach frontier performance: 1. @baseten - novel compaction strategies for analyzing large data rooms. 2. @FireworksAI_HQ - matching frontier performance by using frontier as an advisor. 3. @appliedcompute - improving performance and reducing cost of large scale review tables. 4. @trajectorylabs & @nvidia - sovereign continual learning over client matters. We plan to continue to invest heavily in working with research partners and open sourcing our data, models and research as much as possible. We believe open research in legal will be important to building trust in the frontier ecosystem. We are also scaling our research team. Harvey Labs is our internal research group, responsible for pushing the frontier of legal intelligence and working closely with labs, research partners, and academia to bring the frontier of agent research into Harvey. Labs is run by @nikogrupen and @ItsJulioPereyra - Niko worked on multi-agent RL at Google Brain and Julio clerked and worked in BigLaw. We believe this pairing is crucial for building frontier legal AI systems. Together they have already made significant progress in scaling our data and training efforts. The long term goal of Harvey Labs is to contribute to the research and infrastructure required for the legal industry to create a frontier ecosystem. We believe that the best version of legal super intelligence is one where each law firm, enterprise and government owns their own specialized version. We are hiring for Harvey Labs across the post training, agent and data stack and open to acquiring talented teams / neolabs in this space. If interested please DM me.

Read the full research report: appliedcompute.com/research/conti…

The collapse also shows up in the answers themselves. Under various metrics of intra-prompt diversity, a policy trained with GRPO leads to less diverse responses than a trained with adaptive entropy control. Moreover, we observe that entropy allows response diversity to be tuned, where the optimal entropy target correlates with more response diversity.

Preserving entropy is critical for continued training; in modern post-training recipes, entropy is often a fixed resource that gets exhausted over the course of a training run, making it difficult for the model to improve and learn on new tasks. Adaptive entropy control methods expose entropy as a controllable hyperparameter instead of a side effect. We find that in a continued training setting, GRPO leads to entropy collapse which stalls training performance on subsequent training phases. REPO-R (Petrenko et al., 2026) can hold entropy near a specific target, preventing this collapse, and pushing performance higher in follow-on tasks when the target entropy value is properly tuned.

The workflows that make you different shouldn't run on the same general models everyone else rents. Our co-founder @rhythmrg on when to train your own.

Rhythm Garg @rhythmrg

a week ago

x.com/i/article/2066…

6 23 182 132K 291