Models
Every model Magic Shell can route to, grouped by provider. The list below is generated from src/lib/models.ts at build time, so it always matches what the CLI ships.
Cost legend
Section titled “Cost legend”- Free — no-cost model (may be time-limited)
- Lower-cost — open-weight, flash, or haiku-class model
- Premium — flagship or high-capability model
Use the filter panel to narrow by provider, category, or cost. The msh --models command lists only the models available for your currently configured provider.
Free no-cost model (may be time-limited)Lower-cost open-weight, flash, or haiku-classPremium flagship or high-capability
Filter by provider & category
OpenCode Zen
| Model | ID | Category | Cost | Context | Zen API |
|---|---|---|---|---|---|
DeepSeek V4 Flash (Free) Free DeepSeek coding model with a one-million-token context window. | deepseek-v4-flash-free | fast | Free | 1.0M | openai-compatible |
Qwen 3.8 Flash Open-weight Qwen model for fast coding and agentic work. | qwen3.8-flash | fast | Lower-cost | 1M | openai-compatible |
Nemotron 3.5 Lightning (Free) Current free high-throughput reasoning model on OpenCode Zen. | nemotron-3.5-lightning-free | fast | Free | 262K | openai-compatible |
Nemotron 3 Ultra (Free) Large open reasoning model available free on OpenCode Zen. | nemotron-3-ultra-free | reasoning | Free | 1M | openai-compatible |
MiniMax M3 Current open-weight MiniMax model for efficient agentic tasks. | minimax-m3 | smart | Lower-cost | 1.0M | openai-compatible |
GLM 5.2 Open-weight GLM model for complex coding and tool-use workflows. | glm-5.2 | smart | Lower-cost | 1.0M | openai-compatible |
DeepSeek V4 Pro Current DeepSeek V4 reasoning model for complex agentic tasks. | deepseek-v4-pro | reasoning | Lower-cost | 1.0M | openai-compatible |
Kimi K3 Latest Kimi model for long-context coding and agentic work. | kimi-k3 | reasoning | Premium | 1.0M | openai-compatible |
GPT 5.6 Luna GPT 5.6 Luna from OpenAI's latest GPT family. | gpt-5.6-luna | fast | Lower-cost | 1.1M | openai-responses |
GPT 6 Astra GPT 6 Astra from OpenAI's latest GPT family. | gpt-6-astra | reasoning | Premium | 1.1M | openai-responses |
GPT 6 Luna GPT 6 Luna from OpenAI's latest GPT family. | gpt-6-luna | fast | Lower-cost | 1.1M | openai-responses |
GPT 6 Sol GPT 6 Sol from OpenAI's latest GPT family. | gpt-6-sol | reasoning | Premium | 1.1M | openai-responses |
GPT 6.1 Sol GPT 6.1 Sol from OpenAI's latest GPT family. | gpt-6.1-sol | reasoning | Premium | 1.1M | openai-responses |
GPT 5.6 Terra GPT 5.6 Terra from OpenAI's latest GPT family. | gpt-5.6-terra | smart | Premium | 1.1M | openai-responses |
GPT 5.6 Sol GPT 5.6 Sol from OpenAI's latest GPT family. | gpt-5.6-sol | reasoning | Premium | 1.1M | openai-responses |
Claude Sonnet 5.5 Balanced Claude model for fast coding and agentic tasks. | claude-sonnet-5-5 | smart | Premium | 1M | anthropic |
Claude Opus 5.5 Claude model for agentic coding with always-on adaptive thinking. | claude-opus-5-5 | reasoning | Premium | 1M | anthropic |
Claude Sonnet 5 Latest balanced Claude model for harder coding tasks. | claude-sonnet-5 | smart | Premium | 1M | anthropic |
Claude Opus 5 Latest top-tier Claude model for complex reasoning. | claude-opus-5 | reasoning | Premium | 1M | anthropic |
Claude Fable 5.1 Anthropic's latest model for demanding reasoning and long-running agents. | claude-fable-5-1 | reasoning | Premium | 1M | anthropic |
Claude Fable 5 Anthropic's most capable widely available model for long-running agents. | claude-fable-5 | reasoning | Premium | 1M | anthropic |
| No models match these filters. | |||||
OpenRouter
| Model | ID | Category | Cost | Context |
|---|---|---|---|---|
GPT 6.1 Sol Updated Sol model for coding, agents, and professional work. | openai/gpt-6.1-sol | reasoning | Premium | 1.1M |
Claude Sonnet 5.5 Balanced Claude model for fast coding and agentic tasks. | anthropic/claude-sonnet-5.5 | smart | Premium | 1M |
Qwen 3.8 Flash Open-weight multimodal model for coding and agentic workflows. | qwen/qwen3.8-flash | fast | Lower-cost | 1M |
MiMo V2.6 Flash Xiaomi's open-source mixture-of-experts model with long-context support. | xiaomi/mimo-v2.6-flash | fast | Lower-cost | 1.0M |
GPT 6 Luna Efficient GPT-6 model for focused, high-volume tasks. | openai/gpt-6-luna | fast | Lower-cost | 1.1M |
GPT 6 Sol GPT-6 model for complex coding and agentic workflows. | openai/gpt-6-sol | reasoning | Premium | 1.1M |
Claude Opus 5.5 Claude model for agentic coding with always-on adaptive thinking. | anthropic/claude-opus-5.5 | reasoning | Premium | 1M |
GPT 6 Astra OpenAI's flagship model for complex reasoning and coding. | openai/gpt-6-astra | reasoning | Premium | 1.1M |
DeepSeek V4.1 Flash Efficient open-weight model for coding and reasoning with native vision. | deepseek/deepseek-v4.1-flash | fast | Lower-cost | 1.0M |
Nex N2.5 Mini (Free) Free open-weight model for agentic coding tasks. | nex-agi/nex-n2.5-mini:free | smart | Free | 262K |
Ling 3.0 Tiny (Free) Current fast open-weight model available on OpenRouter's free tier. | inclusionai/ling-3.0-tiny:free | fast | Free | 262K |
North Mini Code (Free) Free open-weight model optimized for coding and terminal tasks. | cohere/north-mini-code:free | fast | Free | 256K |
Nemotron 3.5 Lightning (Free) High-throughput open reasoning model on OpenRouter's free tier. | nvidia/nemotron-3.5-lightning:free | fast | Free | 1M |
Nemotron 3 Ultra (Free) Large open reasoning model available on OpenRouter's free tier. | nvidia/nemotron-3-ultra-550b-a55b:free | reasoning | Free | 1M |
Laguna S 2.1 (Free) Open-weight coding model for long-horizon agentic work. | poolside/laguna-s-2.1:free | reasoning | Free | 262K |
DeepSeek V4 Flash 0731 Latest fast DeepSeek V4 checkpoint with a one-million-token context window. | deepseek/deepseek-v4-flash-0731 | fast | Lower-cost | 1.0M |
MiniMax M3 Current open-weight MiniMax model for efficient agentic tasks. | minimax/minimax-m3 | smart | Lower-cost | 1.0M |
GLM 5.2 Open-weight GLM model for complex coding and tool-use workflows. | z-ai/glm-5.2 | smart | Lower-cost | 1.0M |
GLM 5.3 Latest open-weight GLM model for coding, reasoning, and tool use. | z-ai/glm-5.3 | reasoning | Lower-cost | 1.0M |
GLM 5.3 Flash Open-weight multimodal GLM model optimized for efficient agentic coding. | z-ai/glm-5.3-flash | smart | Lower-cost | 1.3M |
Qwen 3.6 27B Apache-licensed dense Qwen model for coding and reasoning. | qwen/qwen3.6-27b | smart | Lower-cost | 262K |
Qwen 3.8 2.4T A95B Latest open-weight Qwen mixture-of-experts model for tools and reasoning. | qwen/qwen3.8-2.4t-a95b | reasoning | Premium | 1.0M |
DeepSeek V4 Pro 0813 GA release of DeepSeek V4 Pro for complex agentic tasks. | deepseek/deepseek-v4-pro-0813 | reasoning | Lower-cost | 1.0M |
Kimi K3 Latest Kimi model for long-context coding and agentic work. | moonshotai/kimi-k3 | reasoning | Premium | 1.0M |
GPT 5.6 Luna Fast, economical model from OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-luna | fast | Lower-cost | 1.1M |
GPT 5.6 Terra Balanced model from OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-terra | smart | Premium | 1.1M |
GPT 5.6 Sol Most capable model in OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-sol | reasoning | Premium | 1.1M |
Claude Sonnet 5 Latest balanced Claude model for harder coding tasks. | anthropic/claude-sonnet-5 | smart | Premium | 1M |
Claude Opus 5 Latest top-tier Claude model for complex reasoning. | anthropic/claude-opus-5 | reasoning | Premium | 1M |
Claude Fable 5.1 Anthropic's latest model for demanding reasoning and long-running agents. | anthropic/claude-fable-5.1 | reasoning | Premium | 1M |
Claude Fable 5 Anthropic's most capable widely available model for long-running agents. | anthropic/claude-fable-5 | reasoning | Premium | 1M |
| No models match these filters. | ||||
Vercel AI Gateway
| Model | ID | Category | Cost | Context |
|---|---|---|---|---|
GPT 6.1 Sol Updated Sol model for coding, agents, and professional work. | openai/gpt-6.1-sol | reasoning | Premium | 1.1M |
Claude Sonnet 5.5 Balanced Claude model for fast coding and agentic tasks. | anthropic/claude-sonnet-5.5 | smart | Premium | 1M |
MiMo V2.6 Flash Xiaomi's open-source mixture-of-experts model with long-context support. | xiaomi/mimo-v2.6-flash | fast | Lower-cost | 1.0M |
GPT 6 Luna Efficient GPT-6 model for focused, high-volume tasks. | openai/gpt-6-luna | fast | Lower-cost | 1.1M |
GPT 6 Sol GPT-6 model for complex coding and agentic workflows. | openai/gpt-6-sol | reasoning | Premium | 1.1M |
Claude Opus 5.5 Claude model for agentic coding with always-on adaptive thinking. | anthropic/claude-opus-5.5 | reasoning | Premium | 1M |
GPT 6 Astra OpenAI's flagship model for complex reasoning and coding. | openai/gpt-6-astra | reasoning | Premium | 1.1M |
DeepSeek V4.1 Flash Efficient open-weight model for coding and reasoning with native vision. | deepseek/deepseek-v4.1-flash | fast | Lower-cost | 1.0M |
GPT 5.6 Luna Fast, economical model from OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-luna | fast | Lower-cost | 1.1M |
GPT 5.6 Terra Balanced model from OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-terra | smart | Premium | 1.1M |
GPT 5.6 Sol Most capable model in OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-sol | reasoning | Premium | 1.1M |
Claude Sonnet 5 Latest balanced Claude model for harder coding tasks. | anthropic/claude-sonnet-5 | smart | Premium | 1M |
Claude Opus 5 Latest top-tier Claude model for complex reasoning. | anthropic/claude-opus-5 | reasoning | Premium | 1M |
Claude Fable 5.1 Anthropic's latest model for demanding reasoning and long-running agents. | anthropic/claude-fable-5.1 | reasoning | Premium | 1M |
Claude Fable 5 Anthropic's most capable widely available model for long-running agents. | anthropic/claude-fable-5 | reasoning | Premium | 1M |
Qwen 3.8 Flash Open-weight multimodal model for coding and agentic workflows. | alibaba/qwen3.8-flash | fast | Lower-cost | 991K |
DeepSeek V4 Pro 0813 GA release of DeepSeek V4 Pro for complex agentic tasks. | deepseek/deepseek-v4-pro-0813 | reasoning | Lower-cost | 1M |
Qwen 3.8 2.4T A95B Latest open-weight Qwen mixture-of-experts model for tools and reasoning. | alibaba/qwen3.8-2.4t-a95b | reasoning | Premium | 262K |
GLM 5.3 Latest open-weight GLM model for coding, reasoning, and tool use. | zai/glm-5.3 | reasoning | Lower-cost | 1M |
GLM 5.3 Flash Open-weight multimodal GLM model optimized for efficient agentic coding. | zai/glm-5.3-flash | smart | Lower-cost | 1M |
Laguna S 2.1 (Free) Open-weight coding model for long-horizon agentic work. | poolside/laguna-s-2.1-free | reasoning | Free | 256K |
| No models match these filters. | ||||
Cloudflare AI Gateway
| Model | ID | Category | Cost | Context |
|---|---|---|---|---|
Workers AI GPT OSS 120B OpenAI's open-weight reasoning model routed through Cloudflare AI Gateway. | workers-ai/@cf/openai/gpt-oss-120b | reasoning | Lower-cost | 32K |
GPT 6 Luna Efficient GPT-6 model for focused, high-volume tasks. | openai/gpt-6-luna | fast | Lower-cost | 1.1M |
GPT 6 Sol GPT-6 model for complex coding and agentic workflows. | openai/gpt-6-sol | reasoning | Premium | 1.1M |
Claude Opus 5.5 Claude model for agentic coding with always-on adaptive thinking. | anthropic/claude-opus-5-5 | reasoning | Premium | 1M |
GPT 6 Astra OpenAI's flagship model for complex reasoning and coding. | openai/gpt-6-astra | reasoning | Premium | 1.1M |
GPT 5.6 Luna Fast, economical model from OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-luna | fast | Lower-cost | 1.1M |
GPT 5.6 Terra Balanced model from OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-terra | smart | Premium | 1.1M |
GPT 5.6 Sol Most capable model in OpenAI's latest GPT 5.6 family. | openai/gpt-5.6-sol | reasoning | Premium | 1.1M |
Claude Sonnet 5 Latest balanced Claude model for harder coding tasks. | anthropic/claude-sonnet-5 | smart | Premium | 1M |
Claude Opus 5 Latest top-tier Claude model for complex reasoning. | anthropic/claude-opus-5 | reasoning | Premium | 1M |
Claude Fable 5.1 Anthropic's latest model for demanding reasoning and long-running agents. | anthropic/claude-fable-5-1 | reasoning | Premium | 1M |
Claude Fable 5 Anthropic's most capable widely available model for long-running agents. | anthropic/claude-fable-5 | reasoning | Premium | 1M |
DeepSeek V4 Pro 0813 GA release of DeepSeek V4 Pro for complex agentic tasks. | deepseek/deepseek-v4-pro-0813 | reasoning | Lower-cost | 1M |
Qwen 3.8 2.4T A95B Latest open-weight Qwen mixture-of-experts model for tools and reasoning. | alibaba/qwen3.8-2.4t-a95b | reasoning | Premium | 262K |
GLM 5.3 Latest open-weight GLM model for coding, reasoning, and tool use. | zai/glm-5.3 | reasoning | Lower-cost | 1M |
GLM 5.3 Flash Open-weight multimodal GLM model optimized for efficient agentic coding. | zai/glm-5.3-flash | smart | Lower-cost | 1M |
Laguna S 2.1 (Free) Open-weight coding model for long-horizon agentic work. | poolside/laguna-s-2.1-free | reasoning | Free | 256K |
| No models match these filters. | ||||
Cloudflare Workers AI
| Model | ID | Category | Cost | Context |
|---|---|---|---|---|
GLM 5.3 Open-weight flagship model for long-running agentic coding workflows. | @cf/zai-org/glm-5.3 | reasoning | Lower-cost | 1.0M |
GLM 5.3 Flash Open-weight multimodal model for efficient reasoning and agentic coding. | @cf/zai-org/glm-5.3-flash | smart | Lower-cost | 1.3M |
GLM 5.2 Flagship open-weight coding model with function calling and reasoning. | @cf/zai-org/glm-5.2 | reasoning | Lower-cost | 262K |
DeepSeek V4 Flash 0731 Fast open-weight DeepSeek model with a one-million-token context window. | @cf/deepseek-ai/deepseek-v4-flash-0731 | fast | Lower-cost | 1.0M |
DeepSeek V4 Pro 0813 Open-weight DeepSeek model for complex agentic work. | @cf/deepseek-ai/deepseek-v4-pro-0813 | reasoning | Lower-cost | 1.0M |
Qwen 3.8 27B Open-weight vision-language model with reasoning and function calling. | @cf/qwen/qwen3.8-27b | smart | Lower-cost | 262K |
GLM 4.7 Flash Fast multilingual open-weight model optimized for tool calling. | @cf/zai-org/glm-4.7-flash | fast | Lower-cost | 131K |
Llama 3.3 70B Fast Fast open-weight Llama model hosted by Workers AI. | @cf/meta/llama-3.3-70b-instruct-fp8-fast | smart | Lower-cost | 24K |
Kimi K2.7 Code Open-weight coding model with tool calling and structured outputs. | @cf/moonshotai/kimi-k2.7-code | reasoning | Lower-cost | 262K |
Llama 4 Scout Open-weight mixture-of-experts model with function calling. | @cf/meta/llama-4-scout-17b-16e-instruct | smart | Lower-cost | 131K |
GPT OSS 120B OpenAI open-weight reasoning model hosted by Workers AI. | @cf/openai/gpt-oss-120b | reasoning | Lower-cost | 32K |
| No models match these filters. | ||||