Pricing & Limits
Plain pricing. No surprises. Credits that never expire. See full plans and side-by-side comparison on the pricing page.
DEAL4× usage on DeepSeek V4 Pro. Every dollar of credit goes 4× further on this model.through May 31, 2026- Resets monthly. Your request budget refreshes at the start of every billing cycle.
- Top up automatically. Buy extra credits at model cost. They roll over. They never expire.
- Your code stays yours. Command Code never trains on it. Never stores it.
- Pick a plan or top up credits. Head to Studio > Billing.
- Watch every request, in real time. The Usage page in Studio tracks tokens and request history as it happens.
- Cards. Invoices. Auto-reload. Manage them all from the billing portal at Studio > Billing.
For teams that need their code and data to stay on their own infrastructure. Enterprise gives you the security, compliance, and control your organization requires — without compromise.
Talk to our team at support@commandcode.ai to get started.
| Plan | Price/mo | Credits/mo | Included LLM Usage | Models |
|---|---|---|---|---|
| Go | $1 | $10 | ~15K requests | Open-source models only |
| Pro | $15 | $30 | ~25K requests | |
| Max | $100 | $150 | ~110K requests | Open-source + premium models |
| Ultra | $200 | $300 | ~200K requests | Open-source + premium models |
| Plan | Price/mo | Included LLM Usage | Models |
|---|---|---|---|
| Team Pro | $40 | ~35K requests | Open-source + premium models |
| Enterprise | $5,000+ | Custom | Custom pool |
How it works.
- Monthly subscription credits. Reset at the start of every cycle.
- Top up at model cost. Credits roll over. Forever.
- Enterprise? Email support@commandcode.ai for custom terms, SLAs, and dedicated support.
The numbers above are estimates, not ceilings. Your actual usage depends on the models you pick, the length of your prompts, and how long your conversations run.
- A typical request. ~700–1K input tokens, ~125–200 output tokens, plus ~42K-56K cache reads on average. The calculator below has a cache slider to model your own session.
- A range, not a single number. Cheaper models stretch further. On Go, DeepSeek V4 Flash runs ~60K requests with no cache — ~15K once the typical 52K cache reads are included. Our estimates are the mix average across each plan's typical models.
- A balanced mix. Even split across models within each tier. Pro, Max, and Ultra assume ~35% open-source / ~65% premium.
- Long conversations cost more. Every new turn re-reads the full context, so a long-running session burns through credits faster than a fresh one. Use
/clearor start a new session for unrelated tasks to keep things efficient. - Track the real number. The Usage page in Studio always reflects the actual price charged per request.
Usage calculator
Pick a model. Adjust the sliders. See how far your credits go. These are estimates based on typical patterns — your actual mileage may vary.
Command Code runs on the best models from Anthropic, OpenAI, and the open-source community. Each one chosen for what it does best.
Switch any time with /model in interactive mode.
Premium Models
| Model | Best For |
|---|---|
taste-1 | Learns and applies your personal coding style across every session |
| Claude Opus 4.7 | Most intelligent model for agents and coding |
| Claude Opus 4.6 | Deep reasoning, system architecture, and large multi-file refactors |
| Claude Sonnet 4.6 | Day-to-day coding, feature work, and bug fixes |
| Claude Haiku 4.5 | Fast completions, small edits, and quick lookups |
| GPT-5.5 | Latest frontier model for general-purpose reasoning and complex problem solving |
| GPT-5.4 | General-purpose reasoning and complex problem solving |
| GPT-5.4 Mini | Fast, cost-effective everyday coding and lookups |
| GPT-5.3 Codex | Heavy-duty code generation and long coding sessions |
| Gemini 3.5 Flash | Pro-level coding proficiency, parallel agentic execution |
| Gemini 3.1 Flash Lite | High-volume workhorse model with implicit caching |
| Claude Sonnet 4.5 | Previous-generation Sonnet, being phased out |
Open Source Models
| Model | Best For |
|---|---|
| Kimi K2.6 | Long-horizon coding tasks with vision and design |
| Kimi K2.5 | Multimodal frontend coding with long context |
| GLM-5.1 | Long-horizon autonomous coding agents |
| GLM-5 | Multi-mode thinking with long-range planning and memory |
| MiniMax M2.7 | High-capability end-to-end software engineering |
| MiniMax M2.5 | Cross-platform full-stack agentic development |
| DeepSeek V4 Pro | Hybrid-attention long-context reasoning |
| DeepSeek V4 Flash | Fast, cost-efficient reasoning at scale |
| Qwen 3.6 Max Preview | Vibe coding with efficient agent execution |
| Qwen 3.6 Plus | Agentic coding and reasoning at lower cost |
| Step 3.5 Flash | Fast sparse-MoE agentic reasoning at scale |
Plans include usage at model API rates. You pay for what you use, by model and by token. All prices below are per million tokens. Subscriptions include 2-10x more credits for the same cost as API pricing.
Premium Models
| Model | Input | Output | Cache Read | Cache Write |
|---|---|---|---|---|
| Claude Opus 4.7 | $5 | $25 | $0.50 | $6.25 |
| Claude Opus 4.6 | $5 | $25 | $0.50 | $6.25 |
| Claude Sonnet 4.6 | $3 | $15 | $0.30 | $3.75 |
| Claude Haiku 4.5 | $1 | $5 | $0.10 | $1.25 |
| GPT-5.5 | $5 | $30 | $0.50 | - |
| GPT-5.4 | $2.50 | $15 | $0.25 | - |
| GPT-5.4 Mini | $0.75 | $4.50 | $0.075 | - |
| GPT-5.3 Codex | $2 | $8 | $0.50 | - |
| Gemini 3.5 Flash | $1.50 | $9 | $0.15 | - |
| Gemini 3.1 Flash Lite | $0.25 | $1.50 | $0.03 | - |
| Claude Sonnet 4.5 | $3 | $15 | $0.30 | $3.75 |
Open Source Models
| Model | Input | Output | Cache Read | Cache Write |
|---|---|---|---|---|
| Kimi K2.6 | $0.95 | $4.00 | $0.16 | - |
| Kimi K2.5 | $0.60 | $3.00 | $0.10 | - |
| GLM-5.1 | $1.40 | $4.40 | $0.26 | - |
| GLM-5 | $1.00 | $3.20 | $0.20 | - |
| MiniMax M2.7 | $0.30 | $1.20 | $0.06 | - |
| MiniMax M2.5 | $0.27 | $0.95 | $0.03 | - |
| - | ||||
| DeepSeek V4 Flash | $0.14 | $0.28 | $0.01 | - |
| Qwen 3.6 Max Preview | $1.30 | $7.80 | $0.26 | $1.63 |
| Qwen 3.6 Plus | $0.50 | $3.00 | $0.10 | - |
| Step 3.5 Flash | $0.10 | $0.30 | $0.02 | - |
A context window is the maximum span of tokens (text, images, and code) a model can hold in mind at once. Input prompt plus everything it generates back.
Every Command Code session has its own context. As you prompt, read files, and respond, that context fills up. Command Code compresses older messages automatically, so you can keep going without starting over.
Context window size depends on the model. Command Code supports model-specific context windows, with up to 1 million tokens supported on Claude Sonnet 4.6, Claude Opus 4.7, and DeepSeek V4 Pro.
How do I get started?
How do I get started?
Install npm i -g command-code. Sign in. Code. Pick a plan from Studio > Billing or visit the pricing page. Need more? Buy extra credits any time.
What happens when I run out of credits?
What happens when I run out of credits?
Buy more from Studio > Billing, or enable auto top-up. Extra credits are at model cost. They roll over month to month. They never expire.
How do I buy more credits?
How do I buy more credits?
You can upgrade to one of our plans or purchase extra credits at API pricing. Go to Studio > Billing to purchase extra credits manually or enable auto-reload to top up automatically when your balance is low.
Where can I check my usage?
Where can I check my usage?
Visit the Usage page in Studio. It shows per-request cost, token counts, and a full history of your usage. Learn more in the usage docs.
Do you train on my code?
Do you train on my code?
No. Command Code does not train on your code or store your code snippets. Taste processing runs on your codebase and stores learning data in your project and on your local machine only. See our Privacy Policy for details.
Can I switch plans?
Can I switch plans?
Yes. You can upgrade or downgrade anytime from Settings > Billing in Studio. Changes take effect on your next billing cycle.
How does team billing work?
How does team billing work?
Team plans are billed per seat. Credits are pooled at the team level, so power users and occasional users share from the same allocation. Organization admins can manage seats and billing from the org settings.
What is the right plan for me?
What is the right plan for me?
Go for international users. Pro for active agent users. Max for power users running Command Code all day. Ultra for maxed-out daily usage at the highest request volume. Team for shared usage and pooled billing. Enterprise for custom terms, SLAs, and dedicated support.
What are my payment options?
What are my payment options?
Self-serve plans support all major credit and debit cards via Settings > Billing in Studio. For invoice-based billing or wire transfers, contact us at support@commandcode.ai to discuss Enterprise plans.
How does usage-based pricing work?
How does usage-based pricing work?
Every plan includes a set of credits. Usage is charged at model API rates, by model and by token. Run out? Buy more, or enable auto top-up. Credits roll over. They never expire.
How can I see and manage usage in my organization?
How can I see and manage usage in my organization?
Organization admins can view per-seat usage, manage billing, and monitor team-wide consumption from org settings in Studio. Individual and Org usage is available on the Usage page.
Can I buy Command Code from a reseller or third party?
Can I buy Command Code from a reseller or third party?
No. Command Code subscriptions are only sold directly through commandcode.ai. We do not authorize any resellers or third-party sellers. Subscriptions purchased from any other source are unauthorized and may be suspended or terminated. Purchase only through our official website.
Where are models hosted?
Where are models hosted?
Commercial models are hosted by Anthropic, OpenAI, Google, and Azure on their respective US-based infrastructure (EU on demand). Command Code does not train on your code or store code snippets. When processing requests, data is sent to the provider's API and handled per their privacy policy. Taste data is stored locally in your project directory. See our Privacy Policy for details. For more options, contact us at support@commandcode.ai to discuss Enterprise plans.
What about data and privacy for open-source models?
What about data and privacy for open-source models?
The $1 Go plan is meant for international developers. We want coding agents to work for everyone. Open-source models are available globally with infrastructure in the US, EU, and Singapore for reliable access worldwide. For more options, contact us at support@commandcode.ai to discuss Team & Enterprise plans.
Is Command Code open source?
Is Command Code open source?
Currently Command Code is not open source.
Can I opt out of telemetry?
Can I opt out of telemetry?
Yes. See Telemetry for what's collected and how to disable it.
Do you support zero data retention (ZDR)?
Do you support zero data retention (ZDR)?
Yes. 99% of our models route through ZDR-capable upstreams. Run the CLI with CMD_ZDR=1 (for example, CMD_ZDR=1 cmd) to enforce both zero data retention and no prompt training on every request across all plans. A small handful of models don't yet have a ZDR-capable upstream; under CMD_ZDR=1 those requests will fail rather than route through a non-private provider. We're actively adding ZDR coverage to the remaining models. Need ZDR on a specific model that doesn't have it today? Email support@commandcode.ai.
Where can I ask more questions?
Where can I ask more questions?
Join our Discord community to ask questions, share feedback, and connect with other developers. Follow us on 𝕏 @CommandCodeAI. For private inquiries, email us at support@commandcode.ai.
Limited-time price drops on specific models. Apply automatically. No codes. No toggles. Discounts work on every plan, on top-up credits, and auto-expire on the dates below.
DeepSeek V4 Pro: 4x usage
4× usage on DeepSeek V4 Pro. Every dollar of credit goes 4× further on this model.
DeepSeek V4 Pro is 75% off through May 31, 2026.
Because the model's per-token cost is one-quarter of list price for the duration of this deal, every dollar of credit you spend on DeepSeek V4 Pro is worth 4× the requests it normally would be. Practical implications:
- On any subscription (Go, Pro, Max, Ultra, Team Pro): credits routed to DeepSeek V4 Pro stretch 4× further. A $1 Go plan with $10 credits effectively has up to $40 of DeepSeek V4 Pro usage if you stay on this model till.
- On top-up credits: extra credits you purchase from Studio > Billing get the same treatment. Buying $20 of credits and using DeepSeek V4 Pro is equivalent to $80 of usage.
- No code, no toggle: pick
deepseek-v4-profrom/modelin the CLI (or in Studio) and the discounted rate applies automatically. The Usage page reflects the discounted price per request in real time.
Discount expires May 31, 2026 at 15:59 UTC. Past that time the model returns to full list price ($1.74 input / $3.48 output / $0.0145 cache read per 1M tokens).
How deals work.
- Automatic. Pick the discounted model from
/modelin the CLI or in Studio. The rate drops for the length of the deal. - On every plan. Go, Pro, Max, Ultra. Credits routed to the discounted model stretch further on every tier.
- On top-up credits. Extra credits you buy from Studio > Billing get the same discount when spent on the discounted model.
- Real-time billing. The Usage page reflects the discounted price per request as it happens.
- Auto-expires. When a deal ends, the model returns to its full list price across the docs, the calculator, and Studio. Nothing for you to do.