Teriminal-RSI

Terminal-RSI-27B

README

License: apache-2.0

Model details

Table with columns: Property, Value
Property	Value
Release name	Terminal-RSI-27B
Backbone named in the paper	Qwen3.8-27B
Serialized architecture	`Qwen3_5ForConditionalGeneration` (`model_type: qwen3_5`)
Parameters in exported tensors	27,781,427,952, including the packaged vision components
Stored weight dtype	BF16
Text layers / hidden size	64 / 5,120
Configured context capacity	262,144 tokens
Weight format	Hugging Face Safetensors; 13 files referenced by the weight index
License	Apache-2.0; see `LICENSE`

Training approach

The paper describes the retained sequence:

Base → RSR-1 → RPR-1 → RSR-2 → RPR-2

Recursive Self-Rewrite (RSR): Discover verified trajectories across complementary harnesses, rewrite them into runbooks, screen for leakage and compatibility, and execute them afresh. Accepted public trajectories provide actor-only supervised updates; critic weights are carried through the rewrite stage.
Recursive Policy-Refine (RPR): Recalibrate an inherited critic with the actor frozen, then jointly optimize actor and critic using PPO. The initial Warm-up Critic Model is prepared separately from the four critic-only calibration updates at the start of each refinement phase.
Family Dense Reward (FDR): Normalize verifier-measured partial progress across task families. The paper uses FDR in RPR-1 and the ordinary RPR-2 branch.
Success-Prioritized Reward (SPR): Combine a strict-success bonus with bounded partial-progress credit. The paper's retained Long-Reward RPR-2 branch uses SPR.
Turn-Boundary Skip-Observation GAE (TB-GAE): Formulate credit assignment over generated tokens while skipping observation recursion nodes and applying decay across interaction-turn boundaries. This is a methodological contribution, not an independently established component-level performance gain. The paper explicitly notes that deployment of the estimator requires checkpoint/configuration evidence; its documented first phase uses standard GAE with unit discount and trace factors.

Paper-reported evaluation

These results refer to the paper's retained RPR-2 step-18 model; see the checkpoint distinction above.

Table with columns: Benchmark / metric, Base, Paper's retained Terminal-RSI-27B
Benchmark / metric	Base	Paper's retained Terminal-RSI-27B
Terminal-Bench 4.0 — success (%)	0.00	19.84
Terminal-Bench 2.1 — success (%)	51.70	86.52
Terminal-Bench 3.0 — success (%)	0.00	16.95
Terminal-Bench Hard — success (%)	31.00	79.00
Four-suite equal-weight mean success (%)	20.68	50.58
LHTB — mean process reward	0.21	0.37
LHTB — mean strict completions / 46 tasks	0/46	5/46
SWR100 — success (%)	3.00	6.00

Retained-stage scores are arithmetic means over three inference runs (Mean@3). The four-suite mean excludes LHTB and SWR100 and is not task-weighted. LHTB process reward and strict completion are separate metrics.

Evaluation harness

The paper reports:

temperature 0.60, top-p 0.95;
up to 8,192 generated tokens per turn and 96,000 input tokens per call;
structured JSON actions and explicit thinking mode disabled;
linear interaction history with proactive summarization triggered at 32,000 accumulated tokens;
benchmark-specific termination and time limits, with a configured upper bound of 100,000 turns.

Performance depends on the agent harness, tools, prompts, time budget, and verifier. A standalone chat example does not reproduce these agent evaluations. External frontier-model comparisons are omitted here because the paper notes incomplete primary-source/protocol reconciliation.

Loading the model

Use a Transformers version that supports Qwen3_5ForConditionalGeneration, plus PyTorch, Accelerate, and Safetensors. The packaged configuration records the export version as 5.8.0.dev0; configuration loading was checked locally with Transformers 5.17.0. Full model loading and GPU inference were not tested as part of packaging.

python

import torchfrom transformers import AutoTokenizer, Qwen3_5ForConditionalGeneration model_id = "Teriminal-RSI/Terminal-RSI-27B" tokenizer = AutoTokenizer.from_pretrained(model_id)model = Qwen3_5ForConditionalGeneration.from_pretrained(    model_id,    dtype=torch.bfloat16,    device_map="auto",) messages = [    {"role": "system", "content": "You are a terminal assistant. Explain your next action; do not claim to have run commands."},    {"role": "user", "content": "How would you investigate a Python project whose tests fail because a dependency is missing?"},]prompt = tokenizer.apply_chat_template(    messages,    tokenize=False,    add_generation_prompt=True,    enable_thinking=False,)inputs = tokenizer(prompt, return_tensors="pt").to(model.device)with torch.inference_mode():    outputs = model.generate(        **inputs,        max_new_tokens=1024,        do_sample=True,        temperature=0.6,        top_p=0.95,    )print(tokenizer.decode(outputs[0, inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

This example generates text only. Actual terminal execution requires an external agent loop that validates actions, runs commands in a sandbox, returns observations, and checks completion. The BF16 weight files alone occupy approximately 55.56 GB (51.75 GiB); inference additionally requires memory for activations, runtime buffers, and the KV/state cache. Long-context inference may need substantially more memory.

All 13 indexed Safetensors files are required, including model-missing-from-origin-00001.safetensors and model-missing-from-origin-00002.safetensors. These are part of the checkpoint, not optional leftovers.

Intended use and limitations

Intended uses include research on terminal agents, repository repair, code execution, scientific workflow reconstruction, and long-horizon reinforcement learning with executable verification.

Run generated commands in an isolated environment with restricted credentials, network access, and filesystem permissions. Do not give an autonomous agent unrestricted access to production systems.
The model can produce incorrect code, unsafe commands, misleading explanations, or false claims of completion. Use independent tests and protected verifiers rather than relying on self-reported success.
Benchmark gains do not establish general-purpose reliability, safety, or applicability to every repository or task family.
The recursive-stage comparisons are not compute-matched component ablations. The paper separates observed performance from unsupported causal attribution.
No new benchmark runs, full inference validation, or independent safety evaluation were performed for this upload.

中文简介

Terminal-RSI-27B 是基于 Terminal-RSI 论文所述方法基于 Qwen3.8-27B 为基座的终端智能体模型，面向代码仓库检查、依赖修复、科学工作流重建和需要真实执行与验证的多步骤任务。Terminal-RSI 通过递归自重写（RSR）与继承 critic 的递归策略精炼（RPR）交替迭代，结合任务家族进度奖励与优先严格成功的奖励设计，将局部进展转化为可验证的任务完成。

Citation

bibtex

@article{yang2026terminalrsi,  title = {Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine},  author = {Yang, Junyao and Li, Zongxia and Shi, Yucheng and Li, Zhongzhi and Wang, Ruhan and Zhou, Ruiwen and Mi, Haitao and Yan, Shuicheng and Liang, Leowei},  journal = {arXiv preprint arXiv:2610.22192},  year = {2026}}

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

Teriminal-RSI

Model Tree

Base

this model

Input Modalities

TextImageVideo

Output Modalities

Text

Supported Functionality