New in v0.26.11 Command Code supports Gemini 3.5 Flash. Pro-level coding, parallel agentic execution, 1M token context. Live in /model, same hour Google shipped it.
Command Code
Software Development
San Francisco, CA 3,544 followers
Command Code with your taste; the first coding agent that continuously learns your coding preferences over time.
About us
Command Code with your taste; the first coding agent that observes how you write code and adapts to your preferences over time using our `taste-1` meta neuro-symbolic AI model architecture. Code 10x faster. Review 2x quicker. Bugs 5x slashed. Learn more at https://commandcode.ai/launch
- Website
-
https://commandcode.ai
External link for Command Code
- Industry
- Software Development
- Company size
- 11-50 employees
- Headquarters
- San Francisco, CA
- Type
- Privately Held
- Founded
- 2023
Locations
-
Primary
Get directions
2261 Market Street
STE #5698
San Francisco, CA 94114, US
Employees at Command Code
Updates
-
Any frontier model. Any provider. Zero lock-in. • Switch between Sonnet, Opus, GPT-5, and more • Run open models like GLM, Kimi, DeepSeek, and MiniMax • Bring your own provider with full request control • Pick the right model for every task No lock-in. No compromises. Learn more → commandcode.ai/features
-
-
It remembers. So you don't have to. • AGENTS.md for project instructions and architecture context • Automatic checkpoints before every modification • Rewind code, conversations, or both at any time • `--resume` to continue exactly where you left off • Auto-compact keeps long sessions focused and efficient Project memory, built in. Learn more → commandcode.ai/docs
-
-
Hackable. Out of the box. • MCP support for external tools, APIs, and databases • `/skills` for reusable specialized workflows • Custom `/commands` with markdown templates • Sub-agents for isolated research and parallel work • User-level + project-level customization throughout Built to be remixed. Learn more → commandcode.ai/docs
-
-
A mode for every moment. ↳ Interactive chat for real-time coding ↳ Headless mode for CI/CD + automation ↳ Plan mode for architecture analysis ↳ Auto-accept mode for max speed ↳ Background sandboxes for heavy tasks ↳ Resume sessions anytime with `--id` Get started from $1/month. commandcode.ai/pricing
-
-
Your taste. Learned. `npx taste` ↳ Learns from every accept, reject, and edit ↳ Captures your naming, patterns, structure, and style ↳ Writes reusable project-level skills to `.commandcode/taste/` ↳ Push & pull taste with teammates or the community ↳ Import taste from Claude Code, Cursor, or any agent ↳ Distill coding taste from any local path or GitHub repo Build agents that code like you. commandcode.ai/features
-
-
A CLI built for speed. And taste. ↳ Fast interactions ↳ Crisp outputs ↳ Beautiful terminal experience ↳ One-shot setup ↳ Start coding in seconds ↳ No configuration required Built for developers who care about craft. Get started: https://commandcode.ai
-
-
What a day. Command Code broke 100 billion tokens! Command Code was 3rd & 6th most used coding agent¹ this week and probably the most used coding harness in the world for deepseek v4 models. You can literally load complete docs and full dependency's code before writing a single line with 1M context on deepseek. flash is ~100x cheaper than the alternatives, and at this context size it holds up shockingly well. Our $1/mo Go plan has made ai coding with open models accessible everywhere in the world, esp outside the bubble. A billion tokens of deepseek flash for $1 is a bit unbelievable too. ¹ ranked against agents on openrouter. p.s. we're not on openrouter, we use 12 other providers.
-
-
Command Code reposted this
the best products may not win through ads, they may win by making offers users cant ignore, been trying Command Code go plan and its honestly a really good deal for devs, 1$/month gives u 10$ worth of credits for pretty much any model but especially open weight models, and with deepseek already being 75% off the value is even better, the special thing here is the “taste” feature, which is basically a bunch of rules it makes automatically from your usage, for example if u say always do this it will save it as a taste and consider it in every next prompt, also i liked deepseek here more than on hermes, so the harness influences your model experience a lot, now the part i wanna talk about more is the marketing, even if u have good deals with the providers u still most probably lose money giving away 10x the usage, but this will allow u to get a lot of users pretty fast since its just a good deal, so instead of spending a ton of money on marketing u make an offer that people cant ignore and do the marketing for u, and if u look at the innovation adoption curve, this kind of offer is perfect to pull innovators and early adopters in fast, because they are always the first to try something new if the value is obvious, but the real challenge starts after that, getting to the early majority, and that usually takes more than just a crazy deal, it takes a product people actually wanna keep using
-