HotOpenSourcing/OmniCoder-9B

VISIONPricing:Input $0.431 / Cached $0.0862 / Output $1.12Concurrent Unit Cost:1Model Size:9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Mar 25, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

OmniCoder-9B by HotOpenSourcing is a 9-billion parameter coding agent model, fine-tuned on 425,000+ curated agentic coding trajectories from frontier models like Claude Opus 4.6 and GPT-5.4. Built on Qwen3.5-9B's hybrid architecture, it features a 262K native context window and excels at complex software engineering tasks, tool use, and multi-step reasoning. The model demonstrates strong agentic behavior, including error recovery and the use of minimal edit diffs, making it highly effective for advanced code generation and agentic workflows.

Loading preview...

OmniCoder-9B: A Specialized Coding Agent

OmniCoder-9B is a 9-billion parameter coding agent model developed by Tesslate, built upon the Qwen3.5-9B's hybrid architecture. Its core differentiator is its training on over 425,000 curated agentic coding trajectories, derived from advanced models such as Claude Opus 4.6, GPT-5.4, GPT-5.3-Codex, and Gemini 3.1 Pro. This extensive fine-tuning enables OmniCoder-9B to excel in real-world software engineering tasks, tool utilization, and complex multi-step reasoning.

Key Capabilities

  • Advanced Agentic Behavior: Learns from frontier agent traces, demonstrating robust error recovery (e.g., read-before-write patterns, LSP diagnostics) and generating precise edit diffs instead of full rewrites.
  • Hybrid Architecture: Inherits Qwen3.5's Gated Delta Networks for efficient long-context processing, supporting a native 262,144 token context window.
  • Reasoning Chains: Supports <think>...</think> reasoning for effective problem decomposition.
  • Strong Benchmarks: Achieves 83.8% on GPQA Diamond (pass@1) and 90% on AIME 2025 (pass@5), alongside a 61% improvement on Terminal-Bench 2.0 over its base model.

Good For

  • Complex Code Generation: Ideal for developers requiring a model that can handle intricate coding challenges and multi-step problem-solving.
  • Agentic Workflows: Suited for applications involving automated software development, tool use, and interactive coding environments.
  • Error-Resilient Development: Benefits use cases where the model needs to intelligently recover from errors and apply precise code modifications.

Finetunes
1 models