October 28, 2025
.dotcode-1, built on Spec-3.5, offers enterprise-grade code intelligence, dual reasoning, and IDE integration.
Today, we're releasing .dotcode-1, a new coding language model designed for professional software engineering. .dotcode-1 represents the culmination of extensive research in code understanding, generation, and reasoning.
Built on our Spec-3.5 foundation, .dotcode-1 introduces dual-mode operation: a standard mode optimized for rapid code completion and generation, and a thinking mode that engages deep reasoning for complex agentic decisions and algorithmic challenges. This versatility makes it ideal for everything from quick edits to comprehensive system design.
.dotcode-1 is available now through Orbit - AI IDE where it powers context-aware autocomplete, intelligent refactoring, and automated code review. Enterprise deployment options are available through SVECTOR EDS for teams requiring on-premise solutions.
.dotcode-1 operates in two distinct modes, each optimized for different coding scenarios:
Optimized for rapid code generation, completion, and refactoring. Delivers immediate responses with sub-second latency, perfect for real-time IDE integration, quick bug fixes, and iterative development workflows.
Engages extended chain-of-thought reasoning for complex problems. Analyzes architectural trade-offs, evaluates multiple implementation approaches, and provides detailed explanations. Ideal for system design, algorithm optimization, and challenging debugging scenarios.
With native 262K token context, .dotcode-1 maintains coherent understanding across entire codebases. This enables repository-level code analysis, comprehensive refactoring, and maintaining consistency across thousands of files simultaneously.
.dotcode-1 achieves competitive results across coding, agentic, and browser automation benchmarks, demonstrating versatility beyond traditional code completion.
| Benchmark | .dotcode-1 | Claude Sonnet-4 | GPT-4.1 |
|---|---|---|---|
| SWE-bench Verified | 31.2% | 35.5% | 25.3% |
| SWE-bench (500 turns) | 52.8% | 70.4% | — |
| SWE-bench (100 turns) | 53.2% | 68.0% | 48.6% |
| SWE-bench Multilingual | 33.8% | 53.3% | 31.5% |
| Aider-Polyglot | 50.7% | 56.4% | 52.4% |
| Spider2 | 23.8% | 31.1% | 25.6% |
| Benchmark | .dotcode-1 | Claude Sonnet-4 | GPT-4.1 |
|---|---|---|---|
| WebArena | 42.8% | 51.1% | 44.3% |
| Mind2Web | 41.6% | 47.4% | 49.6% |
| Benchmark | .dotcode-1 | Claude Sonnet-4 | GPT-4.1 |
|---|---|---|---|
| BFCL-v3 | 64.5% | 73.3% | 62.9% |
| TAU-Bench Retail | 70.2% | 80.5% | — |
| TAU-Bench Airline | 49.3% | 60.0% | — |
All benchmarks conducted in controlled environments using standardized evaluation protocols. Results represent averaged performance across multiple runs with consistent prompt templates.
.dotcode-1 powers the intelligence layer of Orbit, SVECTOR's AI-native Integrated Development Environment. Unlike traditional code editors with AI plugins, Orbit is designed from the ground up around .dotcode-1's capabilities, creating a seamless experience where the model understands your entire development context.
Orbit maintains continuous awareness of your project structure, open files, git history, and coding patterns. .dotcode-1 uses this context to provide suggestions that align with your architecture and coding style.
Multi-line code completion that understands not just syntax, but intent. Suggests entire functions, handles edge cases, and maintains consistency with existing code patterns across your repository.
Safely refactor code across multiple files with .dotcode-1 analyzing dependencies, updating imports, and maintaining functionality. Handles complex transformations while preserving behavior.
Real-time code review identifying potential bugs, security vulnerabilities, performance issues, and architectural concerns before you commit. Learns from your team's review patterns.
Enterprise Features: Orbit with .dotcode-1 includes enterprise-grade privacy controls, audit logging, policy enforcement, and on-premise deployment options through SVECTOR EDS. All code remains within your infrastructure with no external data transmission.
.dotcode-1 demonstrates strong performance across major programming languages and frameworks:
Systems
Rust, C++, Go, C
Web & Backend
Python, JavaScript, TypeScript, Java
Mobile & Cross-platform
Swift, Kotlin, Dart, React Native
Functional
Haskell, OCaml, Scala, Elixir
Data & ML
Python (NumPy, PyTorch), R, Julia
Specialized
SQL, Shell scripting, YAML, Terraform
With native 262K context window, .dotcode-1 comprehends entire codebases simultaneously. This enables cross-file refactoring, architectural analysis, and maintaining consistency across thousands of files. The model tracks dependencies, understands module relationships, and can reason about system-wide impacts of code changes.
Beyond simple completion, .dotcode-1 generates production-quality code with proper error handling, documentation, and testing:
In thinking mode, .dotcode-1 excels at complex debugging scenarios. It analyzes stack traces, reproduces issues, identifies root causes, and suggests comprehensive fixes. The model understands common anti-patterns, security vulnerabilities, and performance bottlenecks across different languages and frameworks.
.dotcode-1 includes native function calling capabilities, enabling seamless integration with development tools, APIs, and custom workflows. The model can autonomously invoke tools, handle responses, and chain operations to complete complex tasks.
Development Tools
Git operations, build systems, package managers, linters, formatters
API Integration
REST, GraphQL, gRPC endpoints with authentication and error handling
Database Operations
Query execution, schema inspection, data migration, backup/restore
Cloud Services
AWS, Azure, GCP resource management and deployment automation
.dotcode-1 can operate autonomously to complete multi-step development tasks. For example, given a feature request, it can: analyze requirements, design the architecture, implement code across multiple files, write tests, run validation, commit changes with proper messages, and create pull requests with documentation.
As a powerful code generation system, .dotcode-1 includes comprehensive safety measures to prevent misuse and ensure responsible deployment:
Built-in security scanning identifies common vulnerabilities (SQL injection, XSS, CSRF, etc.) and suggests secure alternatives. The model is trained to avoid generating code with known security flaws.
.dotcode-1 refuses to generate malware, exploit code, or systems designed for malicious purposes. Content filters prevent generation of code that could be used for unauthorized access, data theft, or system compromise.
Training data filtering ensures the model doesn't memorize or reproduce copyrighted code verbatim. Generated code is original and doesn't violate open-source licenses or proprietary software terms.
For Orbit - IDE and API deployments, all code remains within your infrastructure. No training on customer code, no external data transmission. Complete audit trails for compliance and security review.
.dotcode-1 underwent multi-stage training combining curated code datasets, synthetic data generation, and reinforcement learning from human feedback (RLHF):
.dotcode-1 is currently available through:
.dotcode-1 represents a significant milestone in AI-assisted software development, combining scale, intelligence, and practical utility. With 30.5 billion parameters, dual reasoning modes, and deep integration into development workflows, it demonstrates how AI can enhance rather than replace human creativity in coding.
As we continue improving .dotcode-1, our focus remains on three priorities: expanding language and framework coverage, deepening reasoning capabilities for complex architectural decisions, and strengthening safety measures to ensure responsible deployment. We're committed to making advanced coding intelligence accessible to developers worldwide while maintaining the highest standards of security and reliability.
Join thousands of developers already using .dotcode-1 to build better software faster. Whether you're prototyping new ideas in Spec Chat, developing production systems in Orbit, or integrating AI into your team's workflow via API, .dotcode-1 is ready to elevate your development experience.
