Baseten

Baseten is high-performance model serving infrastructure that allows teams to deploy open-source models (DeepSeek, Llama 3, Qwen, Moonshot) with sub-second cold starts, dedicated throughput, and enterprise SLAs.

Website: https://www.baseten.co/

---

Getting an API Key

1. Log in to the Baseten Console.
2. Navigate to Settings ➔ API Keys.
3. Create and copy your Baseten API key. Store it securely.

---

Supported Models

PostQode connects to open-source models hosted via the Baseten Model Library or custom deployed Truss endpoints:

Deep Reasoning & Planning

• deepseek-ai/DeepSeek-R1 - (Recommended for Planning) Frontier 671B reasoning model for multi-step logic and system design
• deepseek-ai/DeepSeek-R1-Distill-Qwen-32B - High-efficiency distilled reasoning model
• moonshotai/Kimi-K2-Thinking - Advanced long-context chain-of-thought reasoning

Coding & Agentic Execution

• Qwen/Qwen2.5-Coder-32B-Instruct - (Recommended for Coding) Surgical diffs, refactoring, and tool calling
• deepseek-ai/DeepSeek-V3 - 671B MoE flagship general-purpose and code generation model
• mistralai/Codestral-2501 - High-performance code completion and generation engine

Foundation & General Intelligence

• meta-llama/Llama-3.3-70B-Instruct - Flagship open-weights foundation model
• meta-llama/Llama-3.1-405B-Instruct - Ultra-scale foundation model for complex architectural analysis
• meta-llama/Llama-3.1-8B-Instruct - Lightweight, ultra-fast model for quick completions

---

Configuration in PostQode

1. Open PostQode Settings () and select Baseten as your provider.
2. Enter your Baseten API Key from the Baseten dashboard.
3. Enter your target Model ID or Model Deployment URL (e.g. Qwen/Qwen2.5-Coder-32B-Instruct or your dedicated Truss endpoint URL).
4. Click Save Settings to connect.

---

Capabilities

• Dedicated GPU instances (H100/A100/L4) for low latency.
• Full streaming response and OpenAI-compatible tool calling.
• Dedicated autoscaling and private VPC deployments.