feat: add MiniMax as LLM provider with M3 default model

.github/workflows/cloud-staging-build.yaml

@@ -118,8 +118,6 @@ jobs:
       - name: Setup extension
         run: |
           cd tmp_extension
-          npm install
-          npm run compile:ts
           npm install -g @vscode/vsce@3.3.2
           npm run build
           vsce package
@@ -182,4 +180,4 @@ jobs:
             git commit -m "version $VERSION"
             git push origin staging
           fi
-          cd ..
+          cd ..

CHANGELOG.md

@@ -1,4 +1,9 @@
-#  (2025-08-18)
+#  (2025-08-25)
+
+
+### Reverts
+
+* Revert "Implemented weekend discount" ([734e0c7](https://github.com/Pythagora-io/pythagora-v1/commit/734e0c726b179a45f235a0fd230a6310c77ae740))
 
 
 

README.md

@@ -8,9 +8,9 @@
 
 <div align="center">
 
-[![Discord Follow](https://dcbadge.vercel.app/api/server/HaqXugmxr9?style=flat)](https://discord.gg/HaqXugmxr9)
+[![Discord](https://img.shields.io/badge/Discord-Join%20Us-5865F2?style=social&logo=discord)](https://discord.gg/HaqXugmxr9)
 [![GitHub Repo stars](https://img.shields.io/github/stars/Pythagora-io/gpt-pilot?style=social)](https://github.com/Pythagora-io/gpt-pilot)
-[![Twitter Follow](https://img.shields.io/twitter/follow/HiPythagora?style=social)](https://twitter.com/HiPythagora)
+[![Twitter Follow](https://img.shields.io/twitter/follow/PythagoraAI?style=social)](https://x.com/PythagoraAI)
 
 </div>
 
@@ -28,29 +28,37 @@
 <br>
 
 <div align="center">
-
+   
 ### GPT Pilot doesn't just generate code, it builds apps!
 
 </div>
 
+<div align="center">
+
+This repo is not being maintained anymore.
+
+# Visit [Pythagora.ai](https://www.pythagora.ai/) for more info
+
+</div>
+
 ---
 <div align="center">
 
-[![See it in action](https://i3.ytimg.com/vi/4g-1cPGK0GA/maxresdefault.jpg)](https://youtu.be/4g-1cPGK0GA)
+[![See it in action](https://img.youtube.com/vi/o1nEvwjKziw/0.jpg)]([https://youtu.be/4g-1cPGK0GA](https://www.youtube.com/watch?v=o1nEvwjKziw))
 
-(click to open the video in YouTube) (1:40min)
+(click to open the video in YouTube) (1:04min)
 
 </div>
 
 ---
 
 <div align="center">
 
-<a href="vscode:extension/PythagoraTechnologies.gpt-pilot-vs-code" target="_blank"><img src="https://github.com/Pythagora-io/gpt-pilot/assets/10895136/5792143e-77c7-47dd-ad96-6902be1501cd" alt="Pythagora-io%2Fgpt-pilot | Trendshift" style="width: 185px; height: 55px;" width="185" height="55"/></a>
+<a href="https://marketplace.visualstudio.com/items?itemName=PythagoraTechnologies.pythagora-vs-code" target="_blank"><img src="https://github.com/Pythagora-io/gpt-pilot/assets/10895136/5792143e-77c7-47dd-ad96-6902be1501cd" alt="Pythagora-io%2Fgpt-pilot | Trendshift" style="width: 185px; height: 55px;" width="185" height="55"/></a>
 
 </div>
 
-GPT Pilot is the core technology for the [Pythagora VS Code extension](https://bit.ly/3IeZxp6) that aims to provide **the first real AI developer companion**. Not just an autocomplete or a helper for PR messages but rather a real AI developer that can write full features, debug them, talk to you about issues, ask for review, etc.
+GPT Pilot is the core technology for the [Pythagora VS Code extension](https://marketplace.visualstudio.com/items?itemName=PythagoraTechnologies.pythagora-vs-code) that aims to provide **the first real AI developer companion**. Not just an autocomplete or a helper for PR messages but rather a real AI developer that can write full features, debug them, talk to you about issues, ask for review, etc.
 
 ---
 
@@ -112,7 +120,7 @@ After you have Python and (optionally) PostgreSQL installed, follow these steps:
 5. `pip install -r requirements.txt` (install the dependencies)
 6. `cp example-config.json config.json` (create `config.json` file)
 7. Set your key and other settings in `config.json` file:
-   - LLM Provider (`openai`, `anthropic` or `groq`) key and endpoints (leave `null` for default) (note that Azure and OpenRouter are suppored via the `openai` setting)
+   - LLM Provider (`openai`, `anthropic`, `groq` or `minimax`) key and endpoints (leave `null` for default) (note that Azure and OpenRouter are suppored via the `openai` setting)
    - Your API key (if `null`, will be read from the environment variables)
    - database settings: sqlite is used by default, PostgreSQL should also work
    - optionally update `fs.ignore_paths` and add files or folders which shouldn't be tracked by GPT Pilot in workspace, useful to ignore folders created by compilers

core/agents/frontend.py

@@ -25,7 +25,7 @@
 from core.llm.parser import DescriptiveCodeBlockParser, OptionalCodeBlockParser
 from core.log import get_logger
 from core.telemetry import telemetry
-from core.ui.base import ProjectStage, UISource
+from core.ui.base import ProjectStage
 
 log = get_logger(__name__)
 
@@ -168,10 +168,6 @@ async def iterate_frontend(self):
         if user_input:
             await self.send_message("Errors detected, fixing...")
         else:
-            await self.ui.send_message(
-                "Use code CODE20 and subscribe https://pythagora.ai/pricing",
-                source=UISource("Congratulations", "success"),
-            )
             answer = await self.ask_question(
                 "Do you want to change anything or report a bug?" if frontend_only else FE_CHANGE_REQ,
                 buttons={"yes": "I'm done building the UI"} if not frontend_only else None,

core/agents/spec_writer.py

@@ -11,7 +11,7 @@
 from core.log import get_logger
 from core.telemetry import telemetry
 from core.templates.registry import PROJECT_TEMPLATES
-from core.ui.base import ProjectStage, UISource
+from core.ui.base import ProjectStage
 
 log = get_logger(__name__)
 
@@ -122,9 +122,6 @@ async def initialize_spec_and_project(self) -> AgentResponse:
 
         await self.ui.send_front_logs_headers("specs_0", ["E1 / T1", "Writing Specification", "working"], "")
 
-        await self.ui.send_message(
-            "Use code CODE20 and subscribe https://pythagora.ai/pricing", source=UISource("Congratulations", "success")
-        )
         await self.send_message(
             "## Write specification\n\nPythagora is generating a detailed specification for app based on your input.",
             # project_state_id="setup",

core/config/__init__.py

@@ -78,6 +78,7 @@ class LLMProvider(str, Enum):
     GROQ = "groq"
     LM_STUDIO = "lm-studio"
     AZURE = "azure"
+    MINIMAX = "minimax"
 
 
 class UIAdapter(str, Enum):

core/llm/base.py

@@ -423,6 +423,7 @@ def for_provider(provider: LLMProvider) -> type["BaseLLMClient"]:
         from .anthropic_client import AnthropicClient
         from .azure_client import AzureClient
         from .groq_client import GroqClient
+        from .minimax_client import MiniMaxClient
         from .openai_client import OpenAIClient
         from .relace_client import RelaceClient
 
@@ -436,6 +437,8 @@ def for_provider(provider: LLMProvider) -> type["BaseLLMClient"]:
             return GroqClient
         elif provider == LLMProvider.AZURE:
             return AzureClient
+        elif provider == LLMProvider.MINIMAX:
+            return MiniMaxClient
         else:
             raise ValueError(f"Unsupported LLM provider: {provider.value}")
 

core/llm/minimax_client.py

@@ -0,0 +1,126 @@
+import datetime
+import os
+import re
+from typing import Optional
+
+import tiktoken
+from httpx import Timeout
+from openai import AsyncOpenAI, RateLimitError
+
+from core.config import LLMProvider
+from core.llm.base import BaseLLMClient
+from core.llm.convo import Convo
+from core.log import get_logger
+
+log = get_logger(__name__)
+tokenizer = tiktoken.get_encoding("cl100k_base")
+
+# MiniMax API base URL (international)
+MINIMAX_BASE_URL = "https://api.minimax.io/v1"
+
+# MiniMax temperature must be in (0.0, 1.0], cannot be 0
+MINIMAX_MIN_TEMPERATURE = 0.01
+
+
+class MiniMaxClient(BaseLLMClient):
+    provider = LLMProvider.MINIMAX
+
+    def _init_client(self):
+        api_key = self.config.api_key or os.environ.get("MINIMAX_API_KEY")
+        self.client = AsyncOpenAI(
+            api_key=api_key,
+            base_url=self.config.base_url or MINIMAX_BASE_URL,
+            timeout=Timeout(
+                max(self.config.connect_timeout, self.config.read_timeout),
+                connect=self.config.connect_timeout,
+                read=self.config.read_timeout,
+            ),
+        )
+
+    async def _make_request(
+        self,
+        convo: Convo,
+        temperature: Optional[float] = None,
+        json_mode: bool = False,
+    ) -> tuple[str, int, int]:
+        temp = self.config.temperature if temperature is None else temperature
+        # MiniMax requires temperature > 0
+        if temp <= 0:
+            temp = MINIMAX_MIN_TEMPERATURE
+
+        completion_kwargs = {
+            "model": self.config.model,
+            "messages": convo.messages,
+            "temperature": temp,
+            "stream": True,
+        }
+
+        # MiniMax does not support response_format; skip json_mode
+
+        stream = await self.client.chat.completions.create(**completion_kwargs)
+        response = []
+        prompt_tokens = 0
+        completion_tokens = 0
+
+        async for chunk in stream:
+            if chunk.usage:
+                prompt_tokens += chunk.usage.prompt_tokens
+                completion_tokens += chunk.usage.completion_tokens
+
+            if not chunk.choices:
+                continue
+
+            content = chunk.choices[0].delta.content
+            if not content:
+                continue
+
+            response.append(content)
+            if self.stream_handler:
+                await self.stream_handler(content)
+
+        response_str = "".join(response)
+
+        # Tell the stream handler we're done
+        if self.stream_handler:
+            await self.stream_handler(None)
+
+        if prompt_tokens == 0 and completion_tokens == 0:
+            prompt_tokens = sum(3 + len(tokenizer.encode(msg["content"])) for msg in convo.messages)
+            completion_tokens = len(tokenizer.encode(response_str))
+            log.warning(
+                "MiniMax response did not include token counts, estimating with tiktoken: "
+                f"{prompt_tokens} input tokens, {completion_tokens} output tokens"
+            )
+
+        return response_str, prompt_tokens, completion_tokens
+
+    def rate_limit_sleep(self, err: RateLimitError) -> Optional[datetime.timedelta]:
+        """
+        Handle MiniMax rate limiting using OpenAI-compatible headers.
+        """
+        headers = err.response.headers
+        if "x-ratelimit-remaining-tokens" not in headers:
+            # Check for retry-after header as fallback
+            if "retry-after" in headers:
+                return datetime.timedelta(seconds=int(headers["retry-after"]))
+            return None
+
+        remaining_tokens = headers["x-ratelimit-remaining-tokens"]
+        time_regex = r"(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?"
+        if remaining_tokens == 0:
+            match = re.search(time_regex, headers.get("x-ratelimit-reset-tokens", ""))
+        else:
+            match = re.search(time_regex, headers.get("x-ratelimit-reset-requests", ""))
+
+        if match:
+            hours = int(match.group(1)) if match.group(1) else 0
+            minutes = int(match.group(2)) if match.group(2) else 0
+            seconds = int(match.group(3)) if match.group(3) else 0
+            total_seconds = hours * 3600 + minutes * 60 + seconds
+        else:
+            total_seconds = 5
+
+        return datetime.timedelta(seconds=total_seconds)
+
+
+__all__ = ["MiniMaxClient"]

example-config.json

@@ -27,6 +27,14 @@
         "azure_deployment": "your-azure-deployment-id",
         "api_version": "2024-02-01"
       }
+    },
+    // Example config for MiniMax (see https://platform.minimax.io/docs/api-reference/text-openai-api)
+    // Available models: MiniMax-M3 (default), MiniMax-M2.7, MiniMax-M2.7-highspeed
+    "minimax": {
+      "base_url": "https://api.minimax.io/v1",
+      "api_key": "your-minimax-api-key",
+      "connect_timeout": 60.0,
+      "read_timeout": 20.0
     }
   },
   // Each agent can use a different model or configuration. The default, as before, is GPT4 Turbo

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "pythagora-core"
-version = "2.0.9"
+version = "2.0.10"
 description = "Build complete apps using AI agents"
 authors = ["Leon Ostrez <leon@pythagora.ai>"]
 license = "FSL-1.1-MIT"