Bolt Adds DeepSeek, GLM, and Kimi to Forge Model Picker

Overview

Bolt has officially expanded its Forge model selection capabilities by integrating three prominent AI providers: DeepSeek, GLM, and Kimi. This update significantly broadens the toolkit available to developers using the Bolt platform, offering diverse options for code generation, reasoning, and natural language processing tasks.

In conjunction with the new model additions, Bolt announced a system-wide reset of Forge tokens for all users. This move ensures that every developer starts with a refreshed quota, allowing immediate access to the newly integrated models without waiting for previous cycles to replenish. The announcement was made directly through Bolt’s official social media channels, highlighting the company’s continued commitment to open and accessible AI development tools.

New Model Integrations

The Bolt Forge model picker now includes three distinct AI architectures, each bringing unique strengths to the table for software engineers and automated coding workflows.

DeepSeek Integration

One of the most significant additions is DeepSeek, a Chinese AI company that has gained substantial traction in the developer community. DeepSeek’s models, particularly the DeepSeek-V3 and R1 variants, are known for their high performance in coding benchmarks and logical reasoning, often competing directly with more expensive Western alternatives.

By adding DeepSeek to Forge, Bolt provides users with a cost-effective and highly capable option for complex coding tasks. This is particularly relevant given the growing demand for affordable, high-quality open-weight models that can handle sophisticated programming challenges without incurring prohibitive costs.

GLM and Kimi Arrivals

Alongside DeepSeek, Bolt has also integrated models from GLM (General Language Model) and Kimi. GLM, developed by Zhipu AI, offers a robust suite of models that perform well across a variety of linguistic and technical tasks. These models are recognized for their strong multilingual capabilities and efficient fine-tuning for specific enterprise use cases.

Kimi, another prominent entrant from Moonshot AI, is renowned for its extended context window capabilities. This feature allows Kimi to process and retain information from very long documents or codebases, making it an invaluable asset for tasks involving large-scale refactoring, documentation analysis, or comprehensive code reviews.

The inclusion of these three models reflects a strategic move to diversify the AI backend options available to Bolt users, ensuring that professionals can choose the right tool for their specific technical requirements.

Token Reset and User Impact

Resetting Forge Tokens

Simultaneous with the model update, Bolt executed a reset of Forge tokens for all existing users. This decision was likely driven by the need to standardize access to the newly integrated models and ensure a fair starting point for everyone on the platform.

Token resets are significant because they effectively refresh a user’s capacity to run AI-assisted operations. For developers who may have exhausted their previous allotments, this reset means immediate, unrestricted access to the new DeepSeek, GLM, and Kimi models for a fresh period. It eliminates the friction of waiting for subscription cycles or manual top-ups.

What This Means for Developers

For the average Bolt user, this update translates to greater flexibility and power. Previously, users might have been limited to a narrower set of providers, potentially forcing them to switch platforms if their preferred model was unavailable or too expensive. Now, the choice lies within the same interface.

Cost Efficiency: With options like DeepSeek, users can potentially reduce operational costs while maintaining high-quality output.
Contextual Depth: The addition of Kimi allows for more complex, long-context projects that were previously difficult to manage within strict token limits.
Versatility: GLM’s broad capabilities ensure that whether a task requires precise coding or nuanced language understanding, there is a specialized model ready to assist.

Strategic Implications for Bolt

Expanding the Ecosystem

Bolt’s decision to broaden its Forge model picker signals a competitive stance in the rapidly evolving AI coding assistant market. By not relying on a single provider, Bolt insulates itself from potential rate hikes, outages, or policy changes from any one vendor. This multi-provider strategy is increasingly becoming a standard for serious enterprise tools.

The integration of Chinese AI models like DeepSeek, GLM, and Kimi also indicates a recognition of their technical merit and cost advantages. These models have consistently ranked highly in independent benchmarks for coding and reasoning, making them attractive additions for any developer-focused platform.

Enhancing Developer Productivity

The core value proposition of Bolt remains its ability to turn ideas into functional code. By expanding the underlying model options, Bolt enhances the likelihood that its AI will understand and execute the user’s intent correctly. Different models have different strengths; some excel at syntax generation, while others are better at architectural planning.

Providing access to a wider array of models allows developers to experiment and find the optimal fit for their workflow. A junior developer might prefer a straightforward model for basic scripts, while a senior engineer might leverage Kimi’s long context for managing an entire repository.

Looking Ahead

Adoption and Community Response

As Bolt rolls out these changes, the developer community is expected to respond with interest. The combination of a token reset and new model access is a tangible benefit that encourages adoption and retention. Early adopters are already sharing their experiences with the new models, providing feedback that will likely shape future updates.

The success of this integration will depend on the seamless execution of these models within the Bolt environment. Users expect consistent performance, low latency, and accurate code generation regardless of which provider they select. Bolt will need to monitor these metrics closely to maintain trust.

Future Model Additions

While DeepSeek, GLM, and Kimi represent a significant step forward, industry observers expect Bolt to continue expanding its Forge lineup. The demand for specialized models—such as those tuned specifically for web development, data science, or mobile app creation—remains high.

Potential future integrations could include other open-source models or partnerships with emerging AI labs. The trend toward a multi-model approach suggests that Bolt’s Forge will become an increasingly versatile hub for AI-assisted development, rather than just a conduit for a single provider’s output.

Conclusion

Bolt’s latest update marks a pivotal moment in its evolution as an AI coding platform. By resetting Forge tokens and introducing DeepSeek, GLM, and Kimi, Bolt has empowered its users with more choices, greater cost efficiency, and enhanced capabilities for handling complex development tasks.

This move underscores the importance of flexibility in the AI tools ecosystem. As models continue to diversify in quality and specialty, platforms that offer easy access to multiple options will hold a distinct advantage. For developers using Bolt, the result is a more robust, adaptable, and powerful environment for bringing their digital ideas to life.