1
01.AI
Shares tags: ai
AI Tool
IQuest-Coder-V1 Review
An open-source LLM that generates, tests, and refines multi-file code with 128K-context support.
shipped Apr 19, 2026updated May 27, 2026aifreemium
1IQuest-Coder-V1 is an open-source code-focused LLM developed by IQuest Lab, the AI research arm of Ubiquant.
2It features a native 128K-context window for multi-file code understanding and generation.
3The model family ranges from 7B to 40B parameters, including "Instruct" and "Thinking" variants.
4It achieved a corrected SWE-Bench Verified score of 76.2% and a BigCodeBench score of 49.9%.
IQuest-Coder-V1 at a Glance
</>Embed "Featured on Stork" Badge▼
<a href="https://www.stork.ai/en/iquest-coder-v1" target="_blank" rel="noopener noreferrer"><img src="https://www.stork.ai/api/badge/iquest-coder-v1?style=dark" alt="IQuest-Coder-V1 - Featured on Stork.ai" height="36" /></a>
[](https://www.stork.ai/en/iquest-coder-v1)
overview
What is IQuest-Coder-V1?
IQuest-Coder-V1 is a code-focused large language model tool developed by IQuest Lab that enables developers and software engineers to perform autonomous software engineering, agentic coding, and deep code intelligence. It generates, tests, and refines multi-file code with native 128K-context support. IQuest-Coder-V1 is a family of open-source code-focused large language models (LLMs) developed by IQuest Lab, the AI research arm of Ubiquant, a Chinese quantitative hedge fund. Released in early 2026, it is engineered for deep code understanding and generation, focusing on the dynamic evolution of software logic. The model series ranges from 7B to 40B parameters and includes "Instruct" variants for general assistance and "Thinking" variants for reasoning-driven problem-solving. It utilizes AI innovations like Code-Flow training, reinforcement learning for reasoning, and a Loop recurrent architecture to understand repository evolution and perform iterative self-review.
quick facts
Quick Facts
| Attribute | Value |
|---|---|
| Developer | IQuest Lab (Ubiquant) |
| Business Model | Open Source |
| Pricing | Free (model weights); no API fees |
| Platforms | Local deployment, Cloud deployment, CLI-based agents |
| API Available | No |
| Integrations | CLI-based agent workflows, ClaudeCode, OpenCode |
| Founded | Early 2026 (initial release) |
| HQ | China |
features
Key Features of IQuest-Coder-V1
IQuest-Coder-V1 incorporates several technical innovations and capabilities designed for advanced software engineering tasks, emphasizing deep code intelligence and autonomous operation.
- 1Open-source code-focused large language model family (7B to 40B parameters).
- 2Native 128K-context window for multi-file codebase understanding and long-range dependencies.
- 3Utilizes Code-Flow training paradigm and Loop recurrent architecture for efficient processing.
- 4Performs iterative self-review and understands repository evolution.
- 5Designed for autonomous software engineering and agentic coding tasks.
- 6Generates, tests, and refines multi-file code across various programming languages.
- 7Supports competitive programming and algorithmic problem-solving with deep reasoning capabilities.
- 8Offers "Instruct" variants for general coding assistance and "Thinking" variants for reasoning-driven problem-solving.
- 9Includes preliminary support for HTML and SVG code generation.
- 10Requires lower compute needs for local or cloud deployment compared to larger proprietary models.
use cases
Who Should Use IQuest-Coder-V1?
IQuest-Coder-V1 is primarily designed for professionals and teams requiring deep code intelligence and autonomous software development capabilities, offering specialized assistance across various coding domains.
- 1**Developers & Software Engineers**: For multi-file code generation, testing, refinement, and accelerating end-to-end software engineering from design to debugging.
- 2**Competitive Programmers**: Optimized for deep reasoning, step-by-step problem decomposition, and algorithmic problem-solving in competitive environments.
- 3**AI Builders & Entrepreneurs**: For building automation scripts, APIs, custom workflows, and integrating customizable AI coding assistance into internal tools.
- 4**Teams Seeking Customizable AI**: For local deployment, fine-tuning, and integration into proprietary systems without API fees or vendor lock-in.
- 5**Refactoring & Root-Cause Analysis**: For understanding complex codebases, performing iterative self-review, and identifying issues efficiently.
pricing
IQuest-Coder-V1 Pricing & Plans
IQuest-Coder-V1 is an open-source tool, making its core models freely accessible for both research and commercial use. The model weights are available for download, allowing users to deploy them on their own infrastructure. While the initial data indicated a "freemium" model, the verified information confirms that the core model weights are free to use, with no API fees or licensing costs. Any "freemium" aspect would likely pertain to potential third-party hosted services or enterprise solutions not detailed in the core offering.
- 1**Free**: Access to all IQuest-Coder-V1 model weights (7B, 14B, 40B, and their variants) for local or cloud deployment. No API fees, rate limits, or hidden costs for direct model usage.
- 2**Custom Deployment**: Users incur costs only for their own compute infrastructure (hardware, cloud instances) required to run the models, offering total control over operational expenses.
competitors
IQuest-Coder-V1 vs Competitors
IQuest-Coder-V1 is positioned as a leading open-source code LLM, directly challenging both proprietary and other open-source models in the autonomous software engineering and code intelligence domains by offering high performance with efficient resource utilization.
1
DeepSeek Coder↗
DeepSeek Coder is a series of open-source code language models trained from scratch on a large corpus of code and natural language, excelling in various coding benchmarks.
Like IQuest-Coder-V1, DeepSeek Coder is an open-source, code-focused LLM available in various sizes and is free for research and commercial use. It emphasizes strong performance on coding benchmarks, similar to IQuest-Coder-V1's focus on SWE-Bench and LiveCodeBench.
2
WizardCoder is an open-source code Large Language Model (LLM) optimized on Llama 2, specifically fine-tuned with the Evol-Instruct algorithm to handle complex coding instructions effectively.
Both WizardCoder and IQuest-Coder-V1 are open-source, code-focused LLMs aiming for state-of-the-art performance in coding tasks. WizardCoder focuses on instruction-following through its fine-tuning approach, while IQuest-Coder-V1 uses a 'Code-Flow' training paradigm and offers 'Instruct' and 'Thinking' variants.
3
OpenCode↗
OpenCode is an open-source, terminal-first AI coding agent that is model-agnostic, allowing users to connect various LLM providers (including local models) for autonomous coding workflows.
OpenCode directly competes in the 'agentic coding' and 'autonomous software engineering' space like IQuest-Coder-V1. While IQuest-Coder-V1 is a family of LLMs, OpenCode acts as an agentic harness that can utilize different LLMs, including free tiers, for code generation and execution.
4
GLM-5.1 (Zhipu AI)↗
GLM-5.1 is a flagship open-source LLM specifically designed for agentic engineering and complex, long-horizon software development tasks, excelling in reasoning, coding, and agentic workflows.
Similar to IQuest-Coder-V1, GLM-5.1 is an open-source LLM focused on agentic engineering and software development, with a strong emphasis on reasoning and handling long-horizon tasks. It is designed to improve its outputs through repeated iteration, aligning with autonomous software engineering goals.
❓
Frequently Asked Questions
+What is IQuest-Coder-V1?
IQuest-Coder-V1 is a code-focused large language model tool developed by IQuest Lab that enables developers and software engineers to perform autonomous software engineering, agentic coding, and deep code intelligence. It generates, tests, and refines multi-file code with native 128K-context support.
+Is IQuest-Coder-V1 free?
Yes, IQuest-Coder-V1 is an open-source tool. Its model weights are freely available for download and deployment, incurring no API fees or licensing costs. Users are only responsible for their own compute infrastructure required to run the models.
+What are the main features of IQuest-Coder-V1?
IQuest-Coder-V1 features a native 128K-context window, Code-Flow training, and a Loop recurrent architecture. It supports autonomous software engineering, agentic coding, multi-file code generation, testing, refinement, and competitive programming, with models ranging from 7B to 40B parameters.
+Who should use IQuest-Coder-V1?
IQuest-Coder-V1 is designed for developers, software engineers, competitive programmers, AI builders, and teams seeking customizable AI coding assistance for tasks like multi-file code generation, refactoring, building automation scripts, and agentic software engineering.
+How does IQuest-Coder-V1 compare to alternatives?
IQuest-Coder-V1 competes with proprietary models like Claude Sonnet and GPT-5.1 by achieving comparable benchmark scores (e.g., 76.2% on SWE-Bench Verified) with significantly fewer parameters. Against open-source alternatives like DeepSeek Coder and WizardCoder, it differentiates with its Code-Flow training and Loop architecture, while OpenCode is an agentic harness that can utilize IQuest-Coder-V1 as an underlying LLM.
For builders
This page is doing a job for someone else’s tool.
AI agents read it. Buyers find it. Backlinks accrue. Your tool can have one too — live in 24 hours, indexed by Claude, ChatGPT, and Perplexity, queryable via MCP.