# Command Code Integrates 12 Frontier Models Including MiMo V2.6 Variants

> Platform expands catalog with MiMo V2.6 variants, DeepSeek V4.1 Flash, GPT-6 Luna/Sol, Grok 4.7, Claude Opus 5.5 and Qwen 3.8 Omni Flash, offering per-token pricing and intelligence scores via GOAT/Pro plans and Provider API.

*Published 2026-09-24 · By Marcus Vance*

Command Code is a platform that aggregates frontier AI models from multiple providers with unified pricing and performance metrics.

The recent addition of 12 frontier models to the Command Code platform marks an important development in the accessibility of advanced AI technologies. This expansion allows users to choose from a diverse set of high-performing models without needing to navigate multiple separate services. The platform's model catalog now includes these new entries with specific details on their capabilities, pricing structures, and benchmark scores. Developers can leverage this to optimize their applications by selecting the most suitable model for each task. This strategic expansion is part of a larger trend where platforms like Command Code are becoming essential intermediaries in the AI ecosystem. By offering these models, the platform addresses the growing demand for versatile AI tools that can handle a variety of tasks from natural language processing to multimodal analysis. The per-token pricing model ensures that users only pay for what they use, which is particularly beneficial for variable workload scenarios. Additionally, the intelligence scores provide a quick reference for comparing model capabilities without having to run extensive tests themselves.

In the context of the fast-evolving AI landscape, such aggregators play a crucial role in democratizing access to frontier models. Command Code has built its reputation on providing reliable access and transparent information about each model's performance. The latest update builds on previous offerings by incorporating the newest releases from key players in the industry. This ensures that users have the latest tools at their disposal for research, development, and production use cases. Furthermore, the expansion is aimed at both individual developers and larger organizations that require reliable access to cutting-edge models without the overhead of direct provider relationships. The GOAT and Pro plans provide tiered access that scales with usage, making advanced AI more affordable for a wider audience. The transparent approach also helps users plan their budgets more effectively when scaling projects.

## Which 12 frontier models did Command Code add to its catalog?

The announcement from the Command Code Bluesky account detailed the shipment of 12 new models. These include the MiMo V2.6 Flash, MiMo V2.6 Pro, and MiMo V2.6 Pro UltraSpeed variants from Xiaomi. Additional models are DeepSeek V4.1 Flash with extended usage options, the free Jev model, GPT-6 Luna and GPT-6 Sol from the GPT-6 series, Grok 4.7 from xAI, GLM 5.3 FlashX, Qwen 3.8 Omni Flash from Alibaba, Claude Opus 5.5 from Anthropic, and Step 5 Preview. This comprehensive list reflects a broad coverage of different architectures and strengths. The announcement came shortly after the models page update around September 24, 2026, providing immediate access to users.

Each model brings unique strengths to the platform. For instance, the GPT-6 Luna variant supports a 1.1 million token context window and carries an intelligence score of 37.3 with pricing set at $0.10 per million input tokens and $0.50 per million output tokens. GPT-6 Sol, on the other hand, achieves a higher intelligence score of 47.5. These details are listed directly in the models page for users to review and compare. The inclusion of these models allows for greater flexibility in handling long-context tasks and complex reasoning problems. Users working on document analysis or extended conversations particularly benefit from the large context support.

The Grok 4.7 model from xAI and the Claude Opus 5.5 from Anthropic add further options for users seeking specific performance profiles. Grok 4.7 is listed with an intelligence score of 46.4. Claude Opus 5.5 stands out with the highest score of 57.6 among the new additions and carries pricing of $4.00 for input and $20.00 for output per million tokens. Qwen 3.8 Omni Flash provides multimodal capabilities from Alibaba. The variety ensures that different user needs, from cost-sensitive applications to high-end performance requirements, are met. This range of options supports diverse use cases including coding assistance, creative writing, and data analysis.

## What are the technical specifications of MiMo V2.6 and DeepSeek V4.1 Flash?

The MiMo V2.6 series from Xiaomi was released on September 22, 2026, and includes the Pro variant with 1.02 trillion parameters. This model achieved a score of 46.32 on the Artificial Analysis Intelligence Index, which positioned it as the strongest open-weight model at the time of its launch, surpassing models like Kimi K3 and Qwen3.8 Max. The series also includes Flash and Pro UltraSpeed versions, offering different trade-offs between speed and capability. These models are open-sourced and available through various APIs including the one on Command Code. The open-sourcing aspect allows researchers to study the architecture and potentially fine-tune for specific domains.

DeepSeek V4.1 Flash, released around September 10 to 19, 2026, features a 552 billion parameter mixture of experts architecture. This design allows it to activate only 8 billion parameters per token during the prefill phase and 16 billion during the decode phase. The model includes native multimodal support and a compressed KV cache, which improves efficiency for longer sequences. These technical features make it particularly suitable for applications requiring both high performance and resource efficiency. The compressed KV cache specifically helps in managing memory usage during extended interactions.

The architecture of these models highlights the ongoing trends in AI development toward larger parameter counts combined with efficient activation patterns. For MiMo, the large scale enables high intelligence scores while the variants provide options for different deployment scenarios. DeepSeek's MoE approach reduces the computational cost per token, making it more accessible for frequent use. Command Code's integration brings these technical advantages to a broader user base through its unified interface. The combination of scale and efficiency in these models represents current best practices in frontier model design.

## How do the intelligence scores and pricing compare among the new models?

A comparison of the available metrics reveals significant variation in performance and cost. Claude Opus 5.5 leads with 57.6, followed by Grok 4.7 at 46.4 and MiMo V2.6 Pro at 46.3. GPT-6 Sol scores 47.5, placing it competitively among the group. These scores are based on the platform's intelligence index, which provides a standardized way to evaluate model capabilities across different benchmarks. Users can use these numbers to make informed decisions about which model to use for their specific applications. The scores help quantify the relative strengths in areas such as reasoning, knowledge retrieval, and instruction following.

Comparison of intelligence scores and pricing for selected new models on Command CodeModelIntelligence ScoreInput Price per 1M TokensOutput Price per 1M TokensProviderClaude Opus 5.557.6$4.00$20.00AnthropicMiMo V2.6 Pro46.3VariesVariesXiaomiGrok 4.746.4VariesVariesxAIDeepSeek V4.1 FlashNot listed$0.15$0.60DeepSeekGPT-6 Sol47.5VariesVariesOpenAI-likeQwen 3.8 Omni FlashNot listedVariesVariesAlibaba

The pricing structures vary widely, with DeepSeek V4.1 Flash offering off-peak rates of $0.15 input and $0.60 output per million tokens, making it one of the more affordable options for high-volume usage. In contrast, Claude Opus 5.5 commands premium pricing reflective of its top score. This range allows users to balance performance with cost, selecting premium models for critical tasks and more economical ones for routine operations. The table above summarizes the key data points for quick reference. Off-peak pricing on DeepSeek further incentivizes usage during lower demand periods, optimizing overall expenditure.

## What are the market and stakeholder implications of this expansion?

The addition of these models has several implications for the market. Developers gain access to a wider array of tools without the need for multiple subscriptions, potentially reducing overall costs and complexity. Enterprises can experiment with different models more easily, leading to better optimized solutions for their AI initiatives. Model providers benefit from increased exposure as their models are featured on a popular aggregation platform. This dynamic fosters greater competition and innovation in the frontier models space. The availability through a single platform also accelerates the adoption rate as integration efforts are minimized.

Stakeholders such as researchers and startups particularly benefit from the affordable access through the GOAT and Pro plans. The Provider API further enables integration into larger systems, supporting the development of AI agents and other applications. By lowering the barriers to entry for using these advanced models, Command Code contributes to the broader adoption of AI technology across industries. The transparent listing of scores and prices also promotes informed decision making among users. This transparency reduces the information asymmetry that often exists between model providers and end users.

## How have experts reacted to the model additions on Command Code?

Expert commentary has focused on the practical benefits of integrating these models through a single platform. Ahmad Awais from CommandCode.ai has highlighted the performance potential of models like DeepSeek when properly harnessed. His observations point to the importance of the integration layer in unlocking the full capabilities of frontier models. The addition of extended usage for DeepSeek V4.1 Flash addresses previous limitations in harness quality that may have affected perceived model performance.

> DeepSeek's weakness was often the harness, not the modelAhmad Awais, CommandCode.ai

This perspective underscores the value that Command Code provides by offering optimized access and extended usage options for models like DeepSeek V4.1 Flash. Other reactions from the community have been positive regarding the variety of choices now available. The inclusion of both open-weight and closed models allows for a balanced approach depending on the user's preferences for openness and performance. Community discussions also note the benefit of having multiple MiMo variants for different speed and capability needs.

## What is next for Command Code in the frontier models space?

Looking ahead, Command Code is likely to continue expanding its catalog with additional models as they are released by the various providers. The platform's focus on providing detailed metrics and competitive pricing positions it well for future growth. Users can expect more updates to the models page and possibly new features in the Provider API to support more complex use cases. The emphasis on both performance and affordability is expected to remain a key aspect of the platform's strategy.

The ongoing competition among model developers suggests that more high-performing options will become available, and Command Code is poised to integrate them promptly. This will further enhance the value proposition for its users. Continued updates to intelligence scores and pricing will keep the catalog current. The platform may also introduce new tools for model comparison and selection to assist users in navigating the growing number of options.

- DeepSeek V4.1 Flash released around September 10-19, 2026.
- MiMo V2.6 series released on September 22, 2026.
- Command Code models page updated around September 24, 2026.
- Bluesky announcement of the 12 new models shipped.
- Continued support through GOAT and Pro plans for user access.

## Sources

1. [Lists GPT-6 Luna (1.1M context, 37.3 intelligence, $0.10/$0.50), GPT-6 Sol (47.5 intelligence), Claude Opus 5.5 (57.6 intelligence, $4.00/$20.00), MiMo V2.6 variants, Grok 4.7 (46.4 intelligence), Qwen 3.8 Omni Flash, DeepSeek V4.1 Flash ($0.15/$0.60 off-peak), and 80+ total models with pricing and capabilities.](https://commandcode.ai/models)
2. [Posts announce individual model launches and recap: 'We’ve shipped 12 new models in Command Code' listing Mimo V2.6 Flash/Pro/UltraSpeed, DeepSeek V4.1 Flash, Jev (free), GPT 6 Luna/Sol, Grok 4.7, GLM 5.3 FlashX, Qwen 3.8 Omni Flash, Opus 5.5, Step 5 Preview; also 'GPT 6 Sol and Luna are live', 'Claude Opus 5.5 is live', 'MiMo V2.6 Pro & Flash are live', 'Grok 4.7 is now live'.](https://bsky.app/profile/commandcode.ai)
3. [Announces MiMo-V2.6 series release and open-sourcing: MiMo-V2.6-Pro (1.02T params), MiMo-V2.6-Flash, MiMo-V2.6-Pro-UltraSpeed; available on MiMo API, OpenRouter, etc.; scores and architecture details.](https://mimo.xiaomi.com/mimo-v2-6)
4. [Features statement from Ahmad Awais on DeepSeek performance.](https://www.alcreon.com/podcast-digest/making-deepseek-v4-outperform-opus-4-7-with-taste-ahmadawais-commandcode-ai)
5. [DeepSeek V4.1 Flash activates only 8B parameters per token during prefill and 16B during decode in its 552B MoE architecture.](https://api-docs.deepseek.com/news/news260910)

---
Source: https://aiintelreport.com/frontier-models/command-code-integrates-12-frontier-models
Index: https://aiintelreport.com/llms.txt · Full text: https://aiintelreport.com/llms-full.txt