Skip navigation
Introducing Harness-1, a 20B search agent trained with a state-externalizing harness.
0:00 / 0:56

Introducing Harness-1, a 20B search agent trained with a state-externalizing harness.

Patrick Jiang

@patpcj2.2K followers

Visit on X
2,931 likesDownloadOriginal post

290,598 views

Introducing Harness-1, a 20B search agent trained with a state-externalizing harness. > frontier-level long-horizon search, rivaling Opus-4.6 and outperforming GPT-5.4 > Context-1-level cost and latency > externalizes candidates, evidence, verification, and search history > open-source

47 Comments

Comments from the original post on X (Twitter)

  • Patrick Jiang's profile picture

    @patpcj

    [1/N] I’ve been wondering: maybe search agents are bad at search partly because we make them do all the paperwork in their head. So I tried a simple idea: externalize the search state, then train the model to use that harness. The result is Harness-1: a 20B search agent that can match or even beat much larger frontier AI on hard long-horizon search tasks.

    81 likes

  • Patrick Jiang's profile picture

    @patpcj

    [2/N] The usual search-agent setup is basically: search → read → search → read → keep appending everything to the transcript. At some point the model is not just “searching” anymore. It is also being asked to be a memory system, a note taker, a verifier, and a librarian.

    47 likes

  • Patrick Jiang's profile picture

    @patpcj

    [3/N] This gets especially weird for RL. The final reward can tell you whether the episode worked, but it often does not tell you why it failed. Was it a bad search? Forgotten evidence? Missing verification? Poor curation? Or the agent just losing track of what it had already seen?

    36 likes

  • Patrick Jiang's profile picture

    @patpcj

    [4/N] Harness-1 tries to separate these two jobs. The model still makes the semantic decisions: what to search, what to read, what to keep, what to verify, when to stop. But the harness maintains the recoverable state around those decisions.

    34 likes

  • Patrick Jiang's profile picture

    @patpcj

    [5/N] Concretely, the harness keeps a working memory with: candidate docs, curated evidence, importance tags, search history, evidence links, verification records, dedup/compression, and context-budget markers. So the agent is not just talking to a search box. It is operating over a workspace.

    40 likes

  • Patrick Jiang's profile picture

    @patpcj

    [6/N] I think this changes what RL is actually learning. Instead of training the model to survive a giant append-only transcript, we train it to use a structured search interface: search, curate, revisit, verify, and submit. Much closer to how I’d want a search agent to work.

    37 likes

  • Patrick Jiang's profile picture

    @patpcj

    [7/N] A fun part: this was not trained with a huge amount of task data. Harness-1 uses 899 filtered SFT trajectories and RL on 3,453 queries. The point is not “less data is always enough.” The point is that a lot of the behavioral prior can live in the harness.

    31 likes

  • Patrick Jiang's profile picture

    @patpcj

    [8/N] The result that made me most excited is transfer. Harness-1 improves over Context-1 by +7.9 recall points on source-family benchmarks. But on held-out transfer benchmarks, the gain is +17.0 points. That’s the part that made the idea feel real to me.

    21 likes

  • Patrick Jiang's profile picture

    @patpcj

    [9/N] The ablations were also pretty revealing. When we disable the harness mechanisms, the model does not just lose some information. It changes behavior: more shallow searching, less reading / verification, worse final curation. So the harness is not just engineering glue.

    23 likes

  • Patrick Jiang's profile picture

    @patpcj

    [10/N] My takeaway: for search agents, “the model” is not the whole learning system. The interface matters. The memory layout matters. The action space matters. The harness matters. If we want RL to teach better search behavior, we should probably stop making the model do all the paperwork in its head.

    27 likes

  • Patrick Jiang's profile picture

    @patpcj

    Paper 📄: Code 💻: Model 🤗: HF Paper:

    97 likes

  • Patrick Jiang's profile picture

    @patpcj

    Huge thanks to @trychroma for fully supporting this work, and to @tinkerapi for the training infra!

    27 likes

  • Patrick Jiang's profile picture

    @patpcj

    Huge shoutout to my awesome collaborators @zhiyiscs @HammadTime @kellyhongsn @PatrickXu565299 @SunJiashuo36 !!

    13 likes

  • RG's profile picture

    @rgvrmdya

    Yo this is insanely cool!

    3 likes

  • Pranav's profile picture

    @IamPranavJ

    everyone will fixate on the 20B. the externalizing harness is the more interesting bet. the ablations show it: turn it off and the model searches shallower, verifies less. the scaffold carries the search behavior, not the parameter count. how far does it generalize past search?

    2 likes

  • Samarth Aggarwal's profile picture

    @Samarth_A

    Your product video looks very polished, kudos! Curious what tool you used to create it?

    2 likes

  • Jonathan Chang's profile picture

    @ChangJonathanC

    @kimbochen cool work . Reminds me of the first version of OpenAI deep research where people find out the it can run code during the research

    2 likes

  • Patrick Donohoe's profile picture

    @patrickdonohoe

    Super cool project. I recently was speaking about this at a conference about how smaller models with higher parameter density+ reasoning ability paired with external knowledge stores are the future. Could be interesting to pair this with a web search api!

    2 likes

  • Vikas Tiwari's profile picture

    @pmvikastiwari

    Will it eat up @ExaAILabs ?

    2 likes

  • Anthony 😎🛹's profile picture

    @djabatt

    Interesting direction. It seems we will continue to find new ways to optimize. What i am curious about is where we land. At some point we get a Linux OS and everyone is happy. I assume we get there with this transformer tech coupled with a harness of sorts.

    1 like