mini-swe-agent 1.15.0__tar.gz → 1.16.0__tar.gz

{mini_swe_agent-1.15.0/src/mini_swe_agent.egg-info → mini_swe_agent-1.16.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mini-swe-agent
-Version: 1.15.0
+Version: 1.16.0
 Summary: Nano SWE Agent - A simple AI software engineering agent
 Author-email: Kilian Lieret <kilian.lieret@posteo.de>, "Carlos E. Jimenez" <carlosej@princeton.edu>
 License: MIT License
@@ -66,6 +66,7 @@ Requires-Dist: mike; extra == "dev"
 Requires-Dist: mkdocs-material; extra == "dev"
 Requires-Dist: mkdocs-glightbox; extra == "dev"
 Requires-Dist: mkdocs-redirects; extra == "dev"
+Requires-Dist: portkey-ai; extra == "dev"
 Dynamic: license-file
 
 <div align="center">
@@ -74,6 +75,7 @@ Dynamic: license-file
 
 # The 100 line AI agent that solves GitHub issues & more
 
+📣 [Gemini 3 Pro reaches 74% on SWE-bench verified with mini-swe-agent!](https://x.com/KLieret/status/1991164693839270372)<br/>
 📣 [New blogpost: Randomly switching between GPT-5 and Sonnet 4 boosts performance](https://www.swebench.com/SWE-bench/blog/2025/08/19/mini-roulette/)
 
 [![Docs](https://img.shields.io/badge/Docs-green?style=for-the-badge&logo=materialformkdocs&logoColor=white)](https://mini-swe-agent.com/latest/)
@@ -94,7 +96,7 @@ Here's some details:
 
 - **Minimal**: Just [100 lines of python](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/agents/default.py) (+100 total for [env](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/environments/local.py),
 [model](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/models/litellm_model.py), [script](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/run/hello_world.py)) — no fancy dependencies!
-- **Powerful:** Resolves >70% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
+- **Powerful:** Resolves >74% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
 - **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
 - **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
 - **Tested:** [![Codecov](https://img.shields.io/codecov/c/github/swe-agent/mini-swe-agent?style=flat-square)](https://codecov.io/gh/SWE-agent/mini-swe-agent)

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/README.md

@@ -4,6 +4,7 @@
 
 # The 100 line AI agent that solves GitHub issues & more
 
+📣 [Gemini 3 Pro reaches 74% on SWE-bench verified with mini-swe-agent!](https://x.com/KLieret/status/1991164693839270372)<br/>
 📣 [New blogpost: Randomly switching between GPT-5 and Sonnet 4 boosts performance](https://www.swebench.com/SWE-bench/blog/2025/08/19/mini-roulette/)
 
 [![Docs](https://img.shields.io/badge/Docs-green?style=for-the-badge&logo=materialformkdocs&logoColor=white)](https://mini-swe-agent.com/latest/)
@@ -24,7 +25,7 @@ Here's some details:
 
 - **Minimal**: Just [100 lines of python](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/agents/default.py) (+100 total for [env](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/environments/local.py),
 [model](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/models/litellm_model.py), [script](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/run/hello_world.py)) — no fancy dependencies!
-- **Powerful:** Resolves >70% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
+- **Powerful:** Resolves >74% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
 - **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
 - **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
 - **Tested:** [![Codecov](https://img.shields.io/codecov/c/github/swe-agent/mini-swe-agent?style=flat-square)](https://codecov.io/gh/SWE-agent/mini-swe-agent)

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/pyproject.toml

@@ -65,6 +65,7 @@ dev = [
     "mkdocs-material",
     "mkdocs-glightbox",
     "mkdocs-redirects",
+    "portkey-ai",
 ]
 
 [project.urls]

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0/src/mini_swe_agent.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mini-swe-agent
-Version: 1.15.0
+Version: 1.16.0
 Summary: Nano SWE Agent - A simple AI software engineering agent
 Author-email: Kilian Lieret <kilian.lieret@posteo.de>, "Carlos E. Jimenez" <carlosej@princeton.edu>
 License: MIT License
@@ -66,6 +66,7 @@ Requires-Dist: mike; extra == "dev"
 Requires-Dist: mkdocs-material; extra == "dev"
 Requires-Dist: mkdocs-glightbox; extra == "dev"
 Requires-Dist: mkdocs-redirects; extra == "dev"
+Requires-Dist: portkey-ai; extra == "dev"
 Dynamic: license-file
 
 <div align="center">
@@ -74,6 +75,7 @@ Dynamic: license-file
 
 # The 100 line AI agent that solves GitHub issues & more
 
+📣 [Gemini 3 Pro reaches 74% on SWE-bench verified with mini-swe-agent!](https://x.com/KLieret/status/1991164693839270372)<br/>
 📣 [New blogpost: Randomly switching between GPT-5 and Sonnet 4 boosts performance](https://www.swebench.com/SWE-bench/blog/2025/08/19/mini-roulette/)
 
 [![Docs](https://img.shields.io/badge/Docs-green?style=for-the-badge&logo=materialformkdocs&logoColor=white)](https://mini-swe-agent.com/latest/)
@@ -94,7 +96,7 @@ Here's some details:
 
 - **Minimal**: Just [100 lines of python](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/agents/default.py) (+100 total for [env](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/environments/local.py),
 [model](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/models/litellm_model.py), [script](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/run/hello_world.py)) — no fancy dependencies!
-- **Powerful:** Resolves >70% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
+- **Powerful:** Resolves >74% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
 - **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
 - **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
 - **Tested:** [![Codecov](https://img.shields.io/codecov/c/github/swe-agent/mini-swe-agent?style=flat-square)](https://codecov.io/gh/SWE-agent/mini-swe-agent)

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/mini_swe_agent.egg-info/SOURCES.txt

@@ -24,6 +24,7 @@ src/minisweagent/config/mini_no_temp.yaml
 src/minisweagent/config/extra/__init__.py
 src/minisweagent/config/extra/swebench.yaml
 src/minisweagent/config/extra/swebench_roulette.yaml
+src/minisweagent/config/extra/swebench_xml.yaml
 src/minisweagent/environments/__init__.py
 src/minisweagent/environments/docker.py
 src/minisweagent/environments/local.py
@@ -34,14 +35,18 @@ src/minisweagent/environments/extra/swerex_docker.py
 src/minisweagent/models/__init__.py
 src/minisweagent/models/anthropic.py
 src/minisweagent/models/litellm_model.py
+src/minisweagent/models/litellm_response_api_model.py
 src/minisweagent/models/openrouter_model.py
 src/minisweagent/models/portkey_model.py
+src/minisweagent/models/portkey_response_api_model.py
+src/minisweagent/models/requesty_model.py
 src/minisweagent/models/test_models.py
 src/minisweagent/models/extra/__init__.py
 src/minisweagent/models/extra/roulette.py
 src/minisweagent/models/utils/__init__.py
 src/minisweagent/models/utils/cache_control.py
 src/minisweagent/models/utils/key_per_thread.py
+src/minisweagent/models/utils/openai_utils.py
 src/minisweagent/run/__init__.py
 src/minisweagent/run/github_issue.py
 src/minisweagent/run/hello_world.py

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/mini_swe_agent.egg-info/requires.txt

@@ -25,6 +25,7 @@ mike
 mkdocs-material
 mkdocs-glightbox
 mkdocs-redirects
+portkey-ai
 
 [full]
 mini-swe-agent[dev]

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/minisweagent/__init__.py

@@ -8,7 +8,7 @@ This file provides:
   unless you want the static type checking.
 """
 
-__version__ = "1.15.0"
+__version__ = "1.16.0"
 
 import os
 from pathlib import Path

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/minisweagent/agents/default.py

@@ -2,7 +2,6 @@
 
 import re
 import subprocess
-from collections.abc import Callable
 from dataclasses import asdict, dataclass
 
 from jinja2 import StrictUndefined, Template
@@ -25,6 +24,7 @@ class AgentConfig:
     )
     format_error_template: str = "Please always provide EXACTLY ONE action in triple backticks."
     action_observation_template: str = "Observation: {{output}}"
+    action_regex: str = r"```bash\s*\n(.*?)\n```"
     step_limit: int = 0
     cost_limit: float = 3.0
 
@@ -54,7 +54,7 @@ class LimitsExceeded(TerminatingException):
 
 
 class DefaultAgent:
-    def __init__(self, model: Model, env: Environment, *, config_class: Callable = AgentConfig, **kwargs):
+    def __init__(self, model: Model, env: Environment, *, config_class: type = AgentConfig, **kwargs):
         self.config = config_class(**kwargs)
         self.messages: list[dict] = []
         self.model = model
@@ -106,7 +106,7 @@ class DefaultAgent:
 
     def parse_action(self, response: dict) -> dict:
         """Parse the action from the message. Returns the action."""
-        actions = re.findall(r"```bash\s*\n(.*?)\n```", response["content"], re.DOTALL)
+        actions = re.findall(self.config.action_regex, response["content"], re.DOTALL)
         if len(actions) == 1:
             return {"action": actions[0].strip(), **response}
         raise FormatError(self.render_template(self.config.format_error_template, actions=actions))

mini_swe_agent-1.16.0/src/minisweagent/config/extra/swebench_xml.yaml

@@ -0,0 +1,215 @@
+agent:
+  system_template: |
+    You are a helpful assistant that can interact multiple times with a computer shell to solve programming tasks.
+    Your response must contain exactly ONE bash code block with ONE command (or commands connected with && or ||).
+
+    Include a THOUGHT section before your command where you explain your reasoning process.
+    Format your response as shown in <format_example>.
+
+    <format_example>
+    THOUGHT: Your reasoning and analysis here
+
+    <bash_code>your_command_here</bash_code>
+    </format_example>
+
+    Failure to follow these rules will cause your response to be rejected.
+  instance_template: |
+    <pr_description>
+    Consider the following PR description:
+    {{task}}
+    </pr_description>
+
+    <instructions>
+    # Task Instructions
+
+    ## Overview
+    You're a software engineer interacting continuously with a computer by submitting commands.
+    You'll be helping implement necessary changes to meet requirements in the PR description.
+    Your task is specifically to make changes to non-test files in the current directory in order to fix the issue described in the PR description in a way that is general and consistent with the codebase.
+
+    IMPORTANT: This is an interactive process where you will think and issue ONE command, see its result, then think and issue your next command.
+
+    For each response:
+    1. Include a THOUGHT section explaining your reasoning and what you're trying to accomplish
+    2. Provide exactly ONE bash command to execute
+
+    ## Important Boundaries
+    - MODIFY: Regular source code files in /testbed (this is the working directory for all your subsequent commands)
+    - DO NOT MODIFY: Tests, configuration files (pyproject.toml, setup.cfg, etc.)
+
+    ## Recommended Workflow
+    1. Analyze the codebase by finding and reading relevant files
+    2. Create a script to reproduce the issue
+    3. Edit the source code to resolve the issue
+    4. Verify your fix works by running your script again
+    5. Test edge cases to ensure your fix is robust
+
+    ## Command Execution Rules
+    You are operating in an environment where
+    1. You write a single command
+    2. The system executes that command in a subshell
+    3. You see the result
+    4. You write your next command
+
+    Each response should include:
+    1. A **THOUGHT** section where you explain your reasoning and plan
+    2. A single bash code block with your command
+
+    Format your responses like included within the <format_example> block:
+
+    <format_example>
+    THOUGHT: Here I explain my reasoning process, analysis of the current situation,
+    and what I'm trying to accomplish with the command below.
+
+    <bash_code>your_command_here </bash_code></format_example>
+
+    **CRITICAL REQUIREMENTS:**
+    - Your response SHOULD include a THOUGHT section explaining your reasoning
+    - Your response MUST include EXACTLY ONE bash code block
+    - This bash block MUST contain EXACTLY ONE command (or a set of commands connected with && or ||)
+    - If you include zero or multiple bash blocks, or no command at all, YOUR RESPONSE WILL FAIL
+    - Do NOT try to run multiple independent commands in separate blocks in one response
+    - Directory or environment variable changes are not persistent. Every action is executed in a new subshell.
+    - However, you can prefix any action with `MY_ENV_VAR=MY_VALUE cd /path/to/working/dir && ...` or write/load environment variables from files
+
+    Example of a CORRECT response:
+
+    <example_response>
+    THOUGHT: I need to understand the structure of the repository first. Let me check what files are in the current directory to get a better understanding of the codebase.
+
+    <bash_code>ls -la</bash_code>
+    </example_response>
+
+    Example of an INCORRECT response:
+
+    <example_response>
+    THOUGHT: I need to examine the codebase and then look at a specific file. I'll run multiple commands to do this.
+
+    <bash_code>ls -la</bash_code>
+
+    Now I'll read the file:
+
+    <bash_code>cat file.txt</bash_code>
+    </example_response>
+
+    If you need to run multiple commands, either:
+
+    1. Combine them in one block using && or ||
+
+    <bash_code>command1 && command2 || echo "Error occurred"</bash_code>
+
+    2. Wait for the first command to complete, see its output, then issue the next command in your following response.
+
+    ## Environment Details
+
+    - You have a full Linux shell environment
+    - Always use non-interactive flags (-y, -f) for commands
+    - Avoid interactive tools like vi, nano, or any that require user input
+    - If a command isn't available, you can install it
+
+    ## Useful Command Examples
+
+    ### Create a new file:
+
+    <bash_code>cat <<'EOF' > newfile.py
+    import numpy as np
+    hello = "world"
+    print(hello)
+    EOF</bash_code>
+
+    ### Edit files with sed:
+
+    Replace all occurrences
+
+    <bash_code>sed -i 's/old_string/new_string/g' filename.py</bash_code>
+
+    Replace only first occurrence
+
+    <bash_code>sed -i 's/old_string/new_string/' filename.py</bash_code>
+
+    Replace first occurrence on line 1
+
+    <bash_code>sed -i '1s/old_string/new_string/' filename.py</bash_code>
+
+    Replace all occurrences in lines 1-10
+
+    <bash_code>sed -i '1,10s/old_string/new_string/g' filename.py</bash_code>
+
+    ### View file content:
+
+    View specific lines with numbers
+
+    <bash_code> nl -ba filename.py | sed -n '10,20p'</bash_code>
+
+    Any other command you want to run
+
+    <bash_code>anything</bash_code>
+
+    ## Submission
+
+    When you've completed your work (reading, editing, testing), and cannot make further progress
+    issue exactly the following command:
+
+    <bash_code>echo COMPLETE_TASK_AND_SUBMIT_FINAL_OUTPUT && git add -A && git diff --cached</bash_code>
+
+    This command will submit your work.
+    You cannot continue working (reading, editing, testing) in any way on this task after submitting.
+    </instructions>
+  action_observation_template: |
+    <returncode>{{output.returncode}}</returncode>
+    {% if output.output | length < 10000 -%}
+    <output>
+    {{ output.output -}}
+    </output>
+    {%- else -%}
+    <warning>
+    The output of your last command was too long.
+    Please try a different command that produces less output.
+    If you're looking at a file you can try use head, tail or sed to view a smaller number of lines selectively.
+    If you're using grep or find and it produced too much output, you can use a more selective search pattern.
+    If you really need to see something from the full command's output, you can redirect output to a file and then search in that file.
+    </warning>
+    {%- set elided_chars = output.output | length - 10000 -%}
+    <output_head>
+    {{ output.output[:5000] }}
+    </output_head>
+    <elided_chars>
+    {{ elided_chars }} characters elided
+    </elided_chars>
+    <output_tail>
+    {{ output.output[-5000:] }}
+    </output_tail>
+    {%- endif -%}
+  format_error_template: |
+    Please always provide EXACTLY ONE action in the `<bash_code>` block, found {{actions|length}} actions.
+
+    Please format your action in a `<bash_code>` block as shown in <response_example>.
+
+    <response_example>
+    Here are some thoughts about why you want to perform the action.
+
+    <bash_code>ls -la</bash_code>
+    </response_example>
+
+    If you have completed your assignment, please consult the first message about how to
+    submit your solution (you will not be able to continue working on this task after that).
+  step_limit: 250
+  cost_limit: 3.
+  action_regex: <bash_code>(.*?)</bash_code>
+
+environment:
+  cwd: "/testbed"
+  timeout: 60
+  env:
+    PAGER: cat
+    MANPAGER: cat
+    LESS: -R
+    PIP_PROGRESS_BAR: 'off'
+    TQDM_DISABLE: '1'
+  environment_class: docker
+
+model:
+  model_name: "minimax/minimax-m2"
+  model_class: openrouter
+  model_kwargs:
+    temperature: 0.0

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/minisweagent/environments/docker.py

@@ -34,7 +34,13 @@ class DockerEnvironmentConfig:
 
 
 class DockerEnvironment:
-    def __init__(self, *, config_class: type = DockerEnvironmentConfig, logger: logging.Logger | None = None, **kwargs):
+    def __init__(
+        self,
+        *,
+        config_class: type = DockerEnvironmentConfig,
+        logger: logging.Logger | None = None,
+        **kwargs,
+    ):
         """This class executes bash commands in a Docker container using direct docker commands.
         See `DockerEnvironmentConfig` for keyword arguments.
         """

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/minisweagent/models/__init__.py

@@ -81,8 +81,11 @@ def get_model_name(input_model_name: str | None = None, config: dict | None = No
 _MODEL_CLASS_MAPPING = {
     "anthropic": "minisweagent.models.anthropic.AnthropicModel",
     "litellm": "minisweagent.models.litellm_model.LitellmModel",
+    "litellm_response": "minisweagent.models.litellm_response_api_model.LitellmResponseAPIModel",
     "openrouter": "minisweagent.models.openrouter_model.OpenRouterModel",
     "portkey": "minisweagent.models.portkey_model.PortkeyModel",
+    "portkey_response": "minisweagent.models.portkey_response_api_model.PortkeyResponseAPIModel",
+    "requesty": "minisweagent.models.requesty_model.RequestyModel",
     "deterministic": "minisweagent.models.test_models.DeterministicModel",
 }
 

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/minisweagent/models/extra/roulette.py

@@ -1,5 +1,4 @@
 import random
-from collections.abc import Callable
 from dataclasses import asdict, dataclass
 
 from minisweagent import Model
@@ -14,7 +13,7 @@ class RouletteModelConfig:
 
 
 class RouletteModel:
-    def __init__(self, *, config_class: Callable = RouletteModelConfig, **kwargs):
+    def __init__(self, *, config_class: type = RouletteModelConfig, **kwargs):
         """This "meta"-model randomly selects one of the models at every call"""
         self.config = config_class(**kwargs)
         self.models = [get_model(config=config) for config in self.config.model_kwargs]
@@ -50,7 +49,7 @@ class InterleavingModelConfig:
 
 
 class InterleavingModel(RouletteModel):
-    def __init__(self, *, config_class: Callable = InterleavingModelConfig, **kwargs):
+    def __init__(self, *, config_class: type = InterleavingModelConfig, **kwargs):
         """This "meta"-model alternates between the models in the sequence for every call"""
         super().__init__(config_class=config_class, **kwargs)
 

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/minisweagent/models/litellm_model.py

@@ -1,6 +1,7 @@
 import json
 import logging
 import os
+from collections.abc import Callable
 from dataclasses import asdict, dataclass, field
 from pathlib import Path
 from typing import Any, Literal
@@ -32,7 +33,7 @@ class LitellmModelConfig:
 
 
 class LitellmModel:
-    def __init__(self, *, config_class: type = LitellmModelConfig, **kwargs):
+    def __init__(self, *, config_class: Callable = LitellmModelConfig, **kwargs):
         self.config = config_class(**kwargs)
         self.cost = 0.0
         self.n_calls = 0

mini_swe_agent-1.16.0/src/minisweagent/models/litellm_response_api_model.py

@@ -0,0 +1,80 @@
+import logging
+from collections.abc import Callable
+from dataclasses import dataclass
+
+import litellm
+from tenacity import (
+    before_sleep_log,
+    retry,
+    retry_if_not_exception_type,
+    stop_after_attempt,
+    wait_exponential,
+)
+
+from minisweagent.models.litellm_model import LitellmModel, LitellmModelConfig
+from minisweagent.models.utils.openai_utils import coerce_responses_text
+
+logger = logging.getLogger("litellm_response_api_model")
+
+
+@dataclass
+class LitellmResponseAPIModelConfig(LitellmModelConfig):
+    pass
+
+
+class LitellmResponseAPIModel(LitellmModel):
+    def __init__(self, *, config_class: Callable = LitellmResponseAPIModelConfig, **kwargs):
+        super().__init__(config_class=config_class, **kwargs)
+        self._previous_response_id: str | None = None
+
+    @retry(
+        stop=stop_after_attempt(10),
+        wait=wait_exponential(multiplier=1, min=4, max=60),
+        before_sleep=before_sleep_log(logger, logging.WARNING),
+        retry=retry_if_not_exception_type(
+            (
+                litellm.exceptions.UnsupportedParamsError,
+                litellm.exceptions.NotFoundError,
+                litellm.exceptions.PermissionDeniedError,
+                litellm.exceptions.ContextWindowExceededError,
+                litellm.exceptions.APIError,
+                litellm.exceptions.AuthenticationError,
+                KeyboardInterrupt,
+            )
+        ),
+    )
+    def _query(self, messages: list[dict[str, str]], **kwargs):
+        try:
+            resp = litellm.responses(
+                model=self.config.model_name,
+                input=messages if self._previous_response_id is None else messages[-1:],
+                previous_response_id=self._previous_response_id,
+                **(self.config.model_kwargs | kwargs),
+            )
+            self._previous_response_id = getattr(resp, "id", None)
+            return resp
+        except litellm.exceptions.AuthenticationError as e:
+            e.message += " You can permanently set your API key with `mini-extra config set KEY VALUE`."
+            raise e
+
+    def query(self, messages: list[dict[str, str]], **kwargs) -> dict:
+        response = self._query(messages, **kwargs)
+        print(response)
+        text = coerce_responses_text(response)
+        try:
+            cost = litellm.cost_calculator.completion_cost(response)
+        except Exception as e:
+            logger.critical(
+                f"Error calculating cost for model {self.config.model_name}: {e}. "
+                "Please check the 'Updating the model registry' section in the documentation. "
+                "http://bit.ly/4p31bi4 Still stuck? Please open a github issue for help!"
+            )
+            raise
+        self.n_calls += 1
+        self.cost += cost
+        from minisweagent.models import GLOBAL_MODEL_STATS
+
+        GLOBAL_MODEL_STATS.add(cost)
+        return {
+            "content": text,
+        }

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/minisweagent/models/portkey_model.py

@@ -22,7 +22,9 @@ logger = logging.getLogger("portkey_model")
 try:
     from portkey_ai import Portkey
 except ImportError:
-    Portkey = None
+    raise ImportError(
+        "The portkey-ai package is required to use PortkeyModel. Please install it with: pip install portkey-ai"
+    )
 
 
 @dataclass
@@ -45,12 +47,8 @@ class PortkeyModelConfig:
 
 
 class PortkeyModel:
-    def __init__(self, **kwargs):
-        if Portkey is None:
-            raise ImportError(
-                "The portkey-ai package is required to use PortkeyModel. Please install it with: pip install portkey-ai"
-            )
-        self.config = PortkeyModelConfig(**kwargs)
+    def __init__(self, *, config_class: type = PortkeyModelConfig, **kwargs):
+        self.config = config_class(**kwargs)
         self.cost = 0.0
         self.n_calls = 0
         if self.config.litellm_model_registry and Path(self.config.litellm_model_registry).is_file():

mini_swe_agent-1.16.0/src/minisweagent/models/portkey_response_api_model.py

@@ -0,0 +1,74 @@
+import logging
+import os
+from dataclasses import dataclass
+
+import litellm
+from tenacity import (
+    before_sleep_log,
+    retry,
+    retry_if_not_exception_type,
+    stop_after_attempt,
+    wait_exponential,
+)
+
+from minisweagent.models import GLOBAL_MODEL_STATS
+from minisweagent.models.portkey_model import PortkeyModel, PortkeyModelConfig
+from minisweagent.models.utils.cache_control import set_cache_control
+from minisweagent.models.utils.openai_utils import coerce_responses_text
+
+logger = logging.getLogger("portkey_response_api_model")
+
+
+@dataclass
+class PortkeyResponseAPIModelConfig(PortkeyModelConfig):
+    pass
+
+
+class PortkeyResponseAPIModel(PortkeyModel):
+    def __init__(self, *, config_class: type = PortkeyResponseAPIModelConfig, **kwargs):
+        super().__init__(config_class=config_class, **kwargs)
+        self._previous_response_id: str | None = None
+
+    @retry(
+        stop=stop_after_attempt(int(os.getenv("MSWEA_MODEL_RETRY_STOP_AFTER_ATTEMPT", "10"))),
+        wait=wait_exponential(multiplier=1, min=4, max=60),
+        before_sleep=before_sleep_log(logger, logging.WARNING),
+        retry=retry_if_not_exception_type((KeyboardInterrupt, TypeError, ValueError)),
+    )
+    def _query(self, messages: list[dict[str, str]], **kwargs):
+        input_messages = messages if self._previous_response_id is None else messages[-1:]
+        resp = self.client.responses.create(
+            model=self.config.model_name,
+            input=input_messages,
+            previous_response_id=self._previous_response_id,
+            **(self.config.model_kwargs | kwargs),
+        )
+        self._previous_response_id = getattr(resp, "id", None)
+        return resp
+
+    def query(self, messages: list[dict[str, str]], **kwargs) -> dict:
+        if self.config.set_cache_control:
+            messages = set_cache_control(messages, mode=self.config.set_cache_control)
+        response = self._query(messages, **kwargs)
+        text = coerce_responses_text(response)
+        try:
+            cost = litellm.cost_calculator.completion_cost(response)
+            assert cost > 0.0, f"Cost is not positive: {cost}"
+        except Exception as e:
+            if self.config.cost_tracking != "ignore_errors":
+                raise RuntimeError(
+                    f"Error calculating cost for model {self.config.model_name}: {e}. "
+                    "You can ignore this issue from your config file with cost_tracking: 'ignore_errors' or "
+                    "globally with export MSWEA_COST_TRACKING='ignore_errors' to ignore this error. "
+                ) from e
+            cost = 0.0
+        self.n_calls += 1
+        self.cost += cost
+        GLOBAL_MODEL_STATS.add(cost)
+        return {
+            "content": text,
+            "extra": {
+                "response": response.model_dump() if hasattr(response, "model_dump") else {},
+                "cost": cost,
+            },
+        }

mini_swe_agent-1.16.0/src/minisweagent/models/requesty_model.py

@@ -0,0 +1,119 @@
+import json
+import logging
+import os
+from dataclasses import asdict, dataclass, field
+from typing import Any
+
+import requests
+from tenacity import (
+    before_sleep_log,
+    retry,
+    retry_if_not_exception_type,
+    stop_after_attempt,
+    wait_exponential,
+)
+
+from minisweagent.models import GLOBAL_MODEL_STATS
+
+logger = logging.getLogger("requesty_model")
+
+
+@dataclass
+class RequestyModelConfig:
+    model_name: str
+    model_kwargs: dict[str, Any] = field(default_factory=dict)
+
+
+class RequestyAPIError(Exception):
+    """Custom exception for Requesty API errors."""
+
+    pass
+
+
+class RequestyAuthenticationError(Exception):
+    """Custom exception for Requesty authentication errors."""
+
+    pass
+
+
+class RequestyRateLimitError(Exception):
+    """Custom exception for Requesty rate limit errors."""
+
+    pass
+
+
+class RequestyModel:
+    def __init__(self, **kwargs):
+        self.config = RequestyModelConfig(**kwargs)
+        self.cost = 0.0
+        self.n_calls = 0
+        self._api_url = "https://router.requesty.ai/v1/chat/completions"
+        self._api_key = os.getenv("REQUESTY_API_KEY", "")
+
+    @retry(
+        stop=stop_after_attempt(10),
+        wait=wait_exponential(multiplier=1, min=4, max=60),
+        before_sleep=before_sleep_log(logger, logging.WARNING),
+        retry=retry_if_not_exception_type(
+            (
+                RequestyAuthenticationError,
+                KeyboardInterrupt,
+            )
+        ),
+    )
+    def _query(self, messages: list[dict[str, str]], **kwargs):
+        headers = {
+            "Authorization": f"Bearer {self._api_key}",
+            "Content-Type": "application/json",
+            "HTTP-Referer": "https://github.com/SWE-agent/mini-swe-agent",
+            "X-Title": "mini-swe-agent",
+        }
+
+        payload = {
+            "model": self.config.model_name,
+            "messages": messages,
+            **(self.config.model_kwargs | kwargs),
+        }
+
+        try:
+            response = requests.post(self._api_url, headers=headers, data=json.dumps(payload), timeout=60)
+            response.raise_for_status()
+            return response.json()
+        except requests.exceptions.HTTPError as e:
+            if response.status_code == 401:
+                error_msg = "Authentication failed. You can permanently set your API key with `mini-extra config set REQUESTY_API_KEY YOUR_KEY`."
+                raise RequestyAuthenticationError(error_msg) from e
+            elif response.status_code == 429:
+                raise RequestyRateLimitError("Rate limit exceeded") from e
+            else:
+                raise RequestyAPIError(f"HTTP {response.status_code}: {response.text}") from e
+        except requests.exceptions.RequestException as e:
+            raise RequestyAPIError(f"Request failed: {e}") from e
+
+    def query(self, messages: list[dict[str, str]], **kwargs) -> dict:
+        response = self._query(messages, **kwargs)
+
+        # Extract cost from usage information
+        usage = response.get("usage", {})
+        cost = usage.get("cost", 0.0)
+
+        # If cost is not available, raise an error
+        if cost == 0.0:
+            raise RequestyAPIError(
+                f"No cost information available from Requesty API for model {self.config.model_name}. "
+                "Cost tracking is required but not provided by the API response."
+            )
+
+        self.n_calls += 1
+        self.cost += cost
+        GLOBAL_MODEL_STATS.add(cost)
+
+        return {
+            "content": response["choices"][0]["message"]["content"] or "",
+            "extra": {
+                "response": response,  # already is json
+            },
+        }
+
+    def get_template_vars(self) -> dict[str, Any]:
+        return asdict(self.config) | {"n_model_calls": self.n_calls, "model_cost": self.cost}

mini_swe_agent-1.16.0/src/minisweagent/models/utils/openai_utils.py

@@ -0,0 +1,41 @@
+import logging
+from typing import Any
+
+from openai.types.responses.response_output_message import ResponseOutputMessage
+
+logger = logging.getLogger("openai_utils")
+
+
+def coerce_responses_text(resp: Any) -> str:
+    """Helper to normalize OpenAI Responses API result to text.
+
+    Works with both OpenAI client responses and LiteLLM/Portkey responses.
+    """
+    text = getattr(resp, "output_text", None)
+    if isinstance(text, str) and text:
+        return text
+
+    try:
+        output = []
+        for item in resp.output:
+            if isinstance(item, dict):
+                content = item.get("content", [])
+            elif isinstance(item, ResponseOutputMessage):
+                content = item.content
+            else:
+                continue
+
+            for content_item in content:
+                if isinstance(content_item, dict):
+                    text_val = content_item.get("text")
+                elif hasattr(content_item, "text"):
+                    text_val = content_item.text
+                else:
+                    continue
+
+                if text_val:
+                    output.append(text_val)
+        return "\n\n".join(output) or ""
+    except (AttributeError, IndexError, TypeError):
+        logger.warning(f"Could not extract text from response: {resp}")
+        return ""

{mini_swe_agent-1.15.0 → mini_swe_agent-1.16.0}/src/minisweagent/run/extra/config.py

@@ -35,9 +35,9 @@ Here's a few popular models and the required API keys:
 
 [bold green]anthropic/claude-sonnet-4-5-20250929[/bold green] ([bold green]ANTHROPIC_API_KEY[/bold green])
 [bold green]openai/gpt-5[/bold green] or [bold green]openai/gpt-5-mini[/bold green] ([bold green]OPENAI_API_KEY[/bold green])
-[bold green]gemini/gemini-2.5-pro[/bold green] ([bold green]GEMINI_API_KEY[/bold green])
+[bold green]gemini/gemini-3-pro-preview[/bold green] ([bold green]GEMINI_API_KEY[/bold green])
 
-[bold]Note: Please always include the provider in the model name.[/bold]
+[bold]Note: Please always include the provider (e.g., "openai/") in the model name.[/bold]
 
 [bold yellow]You can leave any setting blank to skip it.[/bold yellow]