Models & pricing

Capy bills model tokens on Capy credits, machine time, and generated images to one balance, in dollars, shared by your organization. You can also run models on your own subscriptions and API keys, on any plan, and Capy never charges for those tokens.

A plan grants credits each month at a discount. You can also add balance directly, at least $5, and turn on auto-reload.

Plans⁠

Annual billing charges 12 months upfront at 20% off and still grants credits monthly.

PlanMonthlyAnnual (per month)Credits per month
Capy Lite$20$16$20 of credits
Capy Pro$100$80$105 of credits (+5%)
Capy Max$200–$1,000$160–$800110% of the price (+10%)
EnterpriseCustomCustomCustom

Capy Max comes in $100 steps from $200 to $1,000, so $500 a month buys $550 of credits. Your own API keys, gateways, and subscriptions work on every plan, the free plan included, with no charge or markup on their tokens. Higher plans allow more cloud machine capacity.

Try Capy Lite for $1. Your first 7 days of monthly Lite cost $1 with a verified card, then $20 a month. Each card and each organization gets one intro.

Unlimited members⁠

Every plan includes unlimited members. There is no per-seat price and no charge for inviting anyone.

Models⁠

Each model appears once in the model picker, named by the company that makes it, such as anthropic/claude-opus-5.

Prices are per 1M tokens on Capy credits. They are each provider's list prices, with no markup. A ✓ in the BYOK column means you can run the model on your own key from that provider.

The ZDR column shows each model's zero data retention status on Capy credits. ZDR means the provider stores no prompts or outputs. ZDR on request means zero data retention is available for your organization; book a call to turn it on. No ZDR means the provider may keep prompts and outputs under its own retention policy. On your own API key or subscription, your provider's terms apply.

ModelIDContextInput (per 1M)Output (per 1M)BYOKZDR
Claude Fable 5.2anthropic/claude-fable-5-21M$8.00 (cached $0.20)$40.00✓ZDR on request
Claude Fable 5.1anthropic/claude-fable-5-11M$10.00 (cached $0.25)$50.00✓ZDR on request
Claude Opus 5.5 (default)anthropic/claude-opus-5-51M$4.00 (cached $0.20)$20.00✓ZDR on request
Claude Sonnet 5.5anthropic/claude-sonnet-5-51M$2.00 (cached $0.10)$10.00✓ZDR on request
Claude Haiku 5.5anthropic/claude-haiku-5-51M≤100K: $0.10 (cached $0.01)
>100K: $0.50 (cached $0.05)
≤100K: $0.50
>100K: $2.50
✓ZDR on request
Claude Fable 5anthropic/claude-fable-51M$10.00 (cached $1.00)$50.00✓ZDR on request
Claude Opus 5anthropic/claude-opus-51M$5.00 (cached $0.50)$25.00✓ZDR on request
Claude Sonnet 5anthropic/claude-sonnet-51M$2.00 (cached $0.20)$10.00✓ZDR on request
Claude Haiku 4.5anthropic/claude-haiku-4-5200K$1.00 (cached $0.10)$5.00✓ZDR on request
Claude Opus 4.8anthropic/claude-opus-4-81M$5.00 (cached $0.50)$25.00✓ZDR on request
Claude Opus 4.7anthropic/claude-opus-4-71M$5.00 (cached $0.50)$25.00✓ZDR on request
Claude Sonnet 4.6anthropic/claude-sonnet-4-61M$3.00 (cached $0.30)$15.00✓ZDR on request
Claude Opus 4.6anthropic/claude-opus-4-61M$5.00 (cached $0.50)$25.00✓ZDR on request
Claude Opus 4.5anthropic/claude-opus-4-5200K$5.00 (cached $0.50)$25.00✓ZDR on request
GPT-6 Astraopenai/gpt-6-astra1.05M≤272K: $10.00 (cached $1.00)
>272K: $20.00 (cached $2.00)
≤272K: $50.00
>272K: $75.00
✓ZDR on request
GPT-6.1 Solopenai/gpt-6.1-sol1.05M≤272K: $2.00 (cached $0.10)
>272K: $4.00 (cached $0.20)
≤272K: $10.00
>272K: $15.00
✓ZDR on request
GPT-6 Solopenai/gpt-6-sol1.05M≤272K: $2.00 (cached $0.20)
>272K: $4.00 (cached $0.40)
≤272K: $10.00
>272K: $15.00
✓ZDR on request
GPT-6 Lunaopenai/gpt-6-luna1.05M≤272K: $0.10 (cached $0.01)
>272K: $0.20 (cached $0.02)
≤272K: $0.50
>272K: $0.75
✓ZDR on request
GPT-5.6 Solopenai/gpt-5.6-sol1.05M≤272K: $4.00 (cached $0.40)
>272K: $8.00 (cached $0.80)
≤272K: $20.00
>272K: $30.00
✓ZDR on request
GPT-6.1 Terraopenai/gpt-6.1-terra1.05M≤272K: $1.00 (cached $0.05)
>272K: $2.00 (cached $0.10)
≤272K: $5.00
>272K: $7.50
✓ZDR on request
GPT-5.6 Terraopenai/gpt-5.6-terra1.05M≤272K: $2.00 (cached $0.20)
>272K: $4.00 (cached $0.40)
≤272K: $12.00
>272K: $18.00
✓ZDR on request
GPT-5.6 Lunaopenai/gpt-5.6-luna1.05M≤272K: $0.20 (cached $0.02)
>272K: $0.40 (cached $0.04)
≤272K: $1.20
>272K: $1.80
✓ZDR on request
GPT-5.5openai/gpt-5.51.05M≤272K: $5.00 (cached $0.50)
>272K: $10.00 (cached $1.00)
≤272K: $30.00
>272K: $45.00
✓ZDR on request
GPT-5.5 Proopenai/gpt-5.5-pro1.05M≤272K: $30.00
>272K: $60.00
≤272K: $180.00
>272K: $270.00
✓ZDR on request
GPT-5.4 Miniopenai/gpt-5.4-mini400K$0.75 (cached $0.075)$4.50✓ZDR on request
GPT-5.4openai/gpt-5.41.05M≤272K: $2.50 (cached $0.25)
>272K: $5.00 (cached $0.50)
≤272K: $15.00
>272K: $22.50
✓ZDR on request
GPT-5.3 Codexopenai/gpt-5.3-codex400K$1.75 (cached $0.175)$14.00✓ZDR on request
Grok 4.7xai/grok-4.7500K<200K: $2.00 (cached $0.50)
≥200K: $4.00 (cached $1.00)
<200K: $6.00
≥200K: $12.00
✓ZDR on request
Grok 4.6xai/grok-4.6500K<200K: $2.00 (cached $0.50)
≥200K: $4.00 (cached $1.00)
<200K: $6.00
≥200K: $12.00
✓ZDR on request
Grok 4.5xai/grok-4.5500K<200K: $2.00 (cached $0.30)
≥200K: $4.00 (cached $0.60)
<200K: $6.00
≥200K: $12.00
✓ZDR on request
Kimi K3.1 Promoonshotai/kimi-k3.1-pro1.05M$2.00 (cached $0.20)$10.00ZDR
Kimi K3moonshotai/kimi-k31.05M$3.00 (cached $0.30)$15.00ZDR
GLM-5.3zai/glm-5.31.05M$1.40 (cached $0.26)$4.40ZDR
GLM-5.3-Flashzai/glm-5.3-flash1.05M$0.15 (cached $0.03)$0.50ZDR
DeepSeek V4.1 Flash 1017deepseek/deepseek-v4.1-flash-10171.05M$0.30 (cached $0.006)$1.20ZDR
DeepSeek V4 Prodeepseek/deepseek-v4-pro1.05M$1.74 (cached $0.145)$3.48ZDR
DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-07311.05M$0.14 (cached $0.028)$0.28ZDR
Gemini 4 Carbongoogle/gemini-4-carbon1.05M$2.00 (cached $0.10)$10.00✓ZDR on request
Gemini 4 Argongoogle/gemini-4-argon1.05M$2.00 (cached $0.10)$10.00✓ZDR on request
Gemini 4 Hydrogengoogle/gemini-4-hydrogen1.05M$1.50 (cached $0.075)$7.50✓ZDR on request
Gemini 3.8 Flashgoogle/gemini-3.8-flash1.05M$0.75 (cached $0.075)$3.75✓ZDR on request
Gemini 3.7 Flashgoogle/gemini-3.7-flash1.05M$0.75 (cached $0.075)$3.75✓ZDR on request
Gemini 3.6 Flashgoogle/gemini-3.6-flash1.05M$1.50 (cached $0.15)$7.50✓ZDR on request
Qwen3.8 Maxqwen/qwen3.8-max1M$2.00 (cached $0.25)$6.00No ZDR
Qwen3.8 Flashqwen/qwen3.8-flash1M$0.15 (cached $0.016)$0.47ZDR
MiniMax M3minimax/minimax-m3524K$0.30 (cached $0.06)$1.20ZDR
MiniMax M3.1 Flashminimax/minimax-m3.1-flash1.05M$0.15 (cached $0.015)$0.60ZDR
MiMo-V2.6 Proxiaomi/mimo-v2.6-pro1.05M$0.435 (cached $0.0036)$0.87ZDR
MiMo-V2.6 Flashxiaomi/mimo-v2.6-flash1.05M$0.14 (cached $0.0028)$0.28ZDR
Hy4tencent/hy41.05M$0.417 (cached $0.021)$1.25ZDR
Hy4 Previewtencent/hy4-preview1.05M$0.834 (cached $0.042)$2.50ZDR
Hy3tencent/hy3262K$0.132 (cached $0.033)$0.528ZDR
LongCat-2.0meituan/longcat-2.01.05M$0.30 (cached $0.006)$1.20No ZDR
Muse Spark 1.3meta/muse-spark-1.31.05M$1.25 (cached $0.15)$4.25ZDR
Muse Glimmer 30Bmeta/muse-glimmer-30b131K$0.30 (cached $0.04)$1.10ZDR
Nemotron 3.5 Lightningnvidia/nemotron-3.5-lightning-30b-a3b262K$0.07 (cached $0.04)$0.20ZDR
Nemotron 3.5 Super VLnvidia/nemotron-3.5-super-vl-120b-a12b262K$0.15 (cached $0.02)$0.50ZDR
Particle-1particle/particle-11.05M$0.00 (cached $0.00)$0.00No ZDR

Subscriptions⁠

A connected Claude, Codex, GitHub Copilot, or SuperGrok subscription runs the models its plan includes, against that plan's limits instead of your balance. Models marked Azure (organization) run on an Azure account your organization connects.

You can connect more than one Claude Code, Codex, or SuperGrok account. Once one is connected, it shows Add another in Add a connection. A thread uses the accounts in your order. GitHub Copilot takes one account per person.

A Claude Pro or Max plan runs the Claude models it includes on the Claude Agent SDK, and Capy's tools, machines, subagents, and pull requests work the same inside it.

To connect a plan, pick Claude Code in Add a connection in Settings → Models. Run claude setup-token on your computer and paste the token it prints. Model usage counts against your plan's limits and is never charged to your Capy balance. Machine time still bills Capy credits.

ModelIDRuns onContext
Claude Fable 5.2anthropic/claude-fable-5-2Claude1M
Claude Fable 5.1anthropic/claude-fable-5-1Claude1M
Claude Opus 5.5anthropic/claude-opus-5-5Claude1M
Claude Sonnet 5.5anthropic/claude-sonnet-5-5Claude1M
Claude Haiku 5.5anthropic/claude-haiku-5-5Claude1M
GPT-6 Astraopenai/gpt-6-astraCodex1.05M
GPT-6.1 Solopenai/gpt-6.1-solCodex1.05M
GPT-6 Solopenai/gpt-6-solCodex1.05M
GPT-6 Lunaopenai/gpt-6-lunaCodex1.05M
GPT-5.6 Solopenai/gpt-5.6-solCodex400K
GPT-6.1 Terraopenai/gpt-6.1-terraCodex1.05M
GPT-5.6 Terraopenai/gpt-5.6-terraCodex400K
GPT-5.6 Lunaopenai/gpt-5.6-lunaCodex400K
GPT-5.5openai/gpt-5.5Codex400K
Claude Fable 5.1anthropic/claude-fable-5-1Copilot1M
Claude Opus 5.5anthropic/claude-opus-5-5Copilot1M
Claude Sonnet 5.5anthropic/claude-sonnet-5-5Copilot1M
Claude Haiku 5.5anthropic/claude-haiku-5-5Copilot1M
Claude Fable 5anthropic/claude-fable-5Copilot1M
Claude Opus 5anthropic/claude-opus-5Copilot1M
Claude Sonnet 5anthropic/claude-sonnet-5Copilot1M
Claude Haiku 4.5anthropic/claude-haiku-4-5Copilot200K
Claude Opus 4.8anthropic/claude-opus-4-8Copilot200K
Claude Opus 4.7anthropic/claude-opus-4-7Copilot200K
Claude Sonnet 4.6anthropic/claude-sonnet-4-6Copilot200K
Claude Opus 4.6anthropic/claude-opus-4-6Copilot200K
Claude Opus 4.5anthropic/claude-opus-4-5Copilot200K
GPT-6 Astraopenai/gpt-6-astraCopilot1.05M
GPT-6.1 Solopenai/gpt-6.1-solCopilot1.05M
GPT-6 Solopenai/gpt-6-solCopilot1.05M
GPT-6 Lunaopenai/gpt-6-lunaCopilot1.05M
GPT-5.6 Solopenai/gpt-5.6-solCopilot1.05M
GPT-6.1 Terraopenai/gpt-6.1-terraCopilot1.05M
GPT-5.6 Terraopenai/gpt-5.6-terraCopilot1.05M
GPT-5.6 Lunaopenai/gpt-5.6-lunaCopilot1.05M
GPT-5.5openai/gpt-5.5Copilot1.05M
GPT-5.4 Miniopenai/gpt-5.4-miniCopilot400K
GPT-5.4openai/gpt-5.4Copilot1.05M
GPT-5.3 Codexopenai/gpt-5.3-codexCopilot400K
Gemini 3.8 Flashgoogle/gemini-3.8-flashCopilot1M
Gemini 3.7 Flashgoogle/gemini-3.7-flashCopilot1M
Gemini 3.6 Flashgoogle/gemini-3.6-flashCopilot1M
Grok 4.7xai/grok-4.7SuperGrok500K
Grok 4.6xai/grok-4.6SuperGrok500K
Grok 4.5xai/grok-4.5SuperGrok500K
Claude Fable 5anthropic/claude-fable-5Azure (organization)1M
GPT-5.6 Solopenai/gpt-5.6-solAzure (organization)1.05M

Fast mode⁠

Models with a Fast toggle use the provider's priority tier at a higher price. If no fast host is available, the request runs and bills at the standard rate. Fast rates per 1M tokens:

ModelFast input (per 1M)Fast output (per 1M)
Claude Opus 5.5$8.00 (cached $0.40)$40.00
Claude Opus 5$10.00 (cached $1.00)$50.00
Claude Opus 4.8$10.00 (cached $1.00)$50.00
GPT-6 Astra$20.00 (cached $2.00)$100.00
GPT-6.1 Sol$4.00 (cached $0.20)$20.00
GPT-6 Sol$4.00 (cached $0.40)$20.00
GPT-6 Luna$0.20 (cached $0.02)$1.00
GPT-5.6 Sol$8.00 (cached $0.80)$40.00
GPT-6.1 Terra$2.00 (cached $0.10)$10.00
GPT-5.6 Terra$4.00 (cached $0.40)$24.00
GPT-5.6 Luna$0.40 (cached $0.04)$2.40
GPT-5.5$12.50 (cached $1.25)$75.00
GPT-5.4 Mini$1.50 (cached $0.15)$9.00
GPT-5.4$5.00 (cached $0.50)$30.00
GPT-5.3 Codex$3.50 (cached $0.35)$28.00
Grok 4.7<200K: $4.00 (cached $1.00)
≥200K: $8.00 (cached $2.00)
<200K: $12.00
≥200K: $24.00
Grok 4.6<200K: $4.00 (cached $1.00)
≥200K: $8.00 (cached $2.00)
<200K: $12.00
≥200K: $24.00
Grok 4.5<200K: $4.00 (cached $0.60)
≥200K: $8.00 (cached $1.20)
<200K: $12.00
≥200K: $24.00

Reasoning effort⁠

Most models offer a reasoning effort from none up to max, and the picker shows the levels each model supports. An unsupported effort fails instead of falling back to another level.

Connections⁠

Every model runs on Capy credits until you connect a subscription or API key that covers it. A thread uses the first connection that covers its model: the owner's connections in order, then the organization's, then Capy credits.

Settings → Models lets you reorder connections and choose the models each connection serves. A connection can be edited, tested, reconnected, removed, or turned off without removal. Admins can also move their own keys to the organization.

Models settings with personal subscriptions and their usage, a personal key, organization connections, and Capy credits in priority orderModels settings with personal subscriptions and their usage, a personal key, organization connections, and Capy credits in priority order
Connections run in order, with Capy credits last.

Add a connection⁠

The add button beside Your connections or Organization connections opens Add a connection, one searchable list of every provider. Picking a provider opens its connect steps in the same sheet. Search matches a provider's name, its host, or a related word, such as gemini for Google or aws for Amazon Bedrock. Every connection works without a plan, and Capy never charges for the tokens it serves.

GroupProviders
SubscriptionsClaude Code, Codex, GitHub Copilot, SuperGrok
GatewaysVercel AI Gateway, OpenRouter, Cloudflare AI Gateway, Ollama Cloud, Neon AI Gateway, Amazon Bedrock, Baseten, Fireworks, Together AI, DeepInfra, NVIDIA, and the coding plans OpenCode Go, Kimi Code, and GLM Coding Plan
API keysAnthropic, OpenAI, xAI, Google, Meta Model API
CustomCustom endpoint, for any OpenAI-compatible server

Subscriptions are personal, so they appear only when you add to Your connections. Claude Code connects with a setup token, and the other subscriptions sign in with a device code. Keys, gateways, coding plans, and custom endpoints can belong to you or the organization. If no provider matches your search, Use a custom endpoint opens the custom endpoint form.

To connect your Codex subscription:

  1. For a personal account, open ChatGPT Settings → Security and turn on Enable device code authorization for Codex. For a managed workspace, ask your workspace owner to allow device-code sign-in.
  2. In Capy, open Add a connection, pick Codex, and copy the code.
  3. Open Verification page, sign in to ChatGPT, and select your workspace if prompted.
  4. Enter the code and complete the approval in OpenAI. Capy finishes the connection once you approve.

If the switch is missing or unavailable in your workspace, ask your ChatGPT workspace owner to turn on Enable device code authentication for Codex CLI. The control is under Workspace settings → Permissions & roles → Codex Local. Some workspaces label this section Codex and Work Local. After changing the relevant setting, start a new connection in Capy. See OpenAI's sign-in help and device-code sign-in instructions.

Add a connection sheet with a provider search field and the Subscriptions, Gateways, API keys, and Custom groupsAdd a connection sheet with a provider search field and the Subscriptions, Gateways, API keys, and Custom groups
Search for a provider, or pick one from its group.

For a key, Check access lists the available models before you connect. You can exclude models from the connection. Anthropic, OpenAI, xAI, and Google keys also cover that company's new models as they come out.

xAI key form with a passed access check and the three xAI models it servesxAI key form with a passed access check and the three xAI models it serves
Check which models your key can run.

Model picker⁠

The model picker shows a subscription icon, a key badge, or no badge for Capy credits. The logo's tooltip names the connection, such as Runs on your Codex subscription. The thread's cost breakdown lists subscription and key usage under On your keys, not charged.

Model picker search results for GPT-6 models with connection badgesModel picker search results for GPT-6 models with connection badges
Search for a model and check what it runs on.

When a connection runs out⁠

A subscription is used up when it hits its plan's limit, until the limit resets. A brief rate limit doesn't count. A connection can also be turned off, signed out, or rejected. When the first connection can't run a model for any of these reasons, the thread follows the run-out setting on the Models page. An admin sets it for the organization and can let members set it for their own threads.

SettingNext in line is another subscription or keyNext in line is Capy credits
Ask meStops and asksStops and asks
Ask before Capy credits (default)Moves on, with a note in the threadStops and asks
Keep goingMoves on, with a note in the threadMoves on, with a note in the thread

When the thread moves on, it posts a note such as Codex is used up until 11:08 PM. Running on OpenAI team key. It goes back to the earlier connection once that connection works again.

When the thread stops, it shows a card. Wait holds the thread until the reset. Use continues on the next connection, and the thread keeps going from then on, Capy credits included. Only the thread's owner can answer, in the app, in Slack, or with /wait and /continue in the CLI. If no connection can run the model, the thread stops.

Thread card with options to wait for a subscription reset or continue on Capy creditsThread card with options to wait for a subscription reset or continue on Capy credits
Wait for the reset or use the next connection.

A key whose provider account has no credit left doesn't follow this setting. The message fails with the provider's error until you top up that account.

Organization settings⁠

Admins manage the organization's connections and these settings on the Models page. Members can see them but not change them.

  • Members can connect their own subscriptions and keys is on by default. When it's off, every thread runs on the organization's connections.
  • Members can choose what happens when a subscription or key runs out is off by default.
  • Only a named key, such as Only OpenAI team key, keeps a model on the organization's keys, never a member's connection or Capy credits.
  • Turn off removes the model from every picker. Automations pinned to that model fail until you pick another one.
  • Compact at sets how full a thread's context gets before the thread summarizes its history.

The last three settings are in each model's menu in the Models table.

Usage billed to credits⁠

A connected subscription or key pays only for the model tokens it serves. Everything else draws from your organization's balance, so a thread on a subscription still stops when the balance is empty.

UsageBilled to Capy credits
Model tokens on a connected subscription or keyNo, on every plan, with no markup. They count against the subscription's limits or bill to your provider account
Model tokens on any model your connections don't serveYes, at the model prices
Machine time for every thread on a cloud machine: your threads, Captains and their crews, subagents on their own machine, automation runs, and reviewsYes, at the machine prices
Images the agent generatesYes
Model tokens for reviewsYes, never on your subscriptions or keys. A higher review effort in Settings → Review costs more per review
Thread titles and the automation run-when checkNo
Threads that run on your own computerNo machine time. Model tokens follow your connections

A sponsored allowance pays for the work it covers before the balance does.

Run without Capy credits⁠

In the Capy desktop app, pick your computer under Machine and choose a model your subscription serves. That thread spends no Capy credits unless it uses a model on Capy credits or generates images.

Keep a model on your subscription⁠

If a thread bills Capy credits for a model you expected your subscription to cover, check these causes:

  • Your subscription's plan doesn't include the model, so the picker shows no icon for it. The Models table shows what serves each model in threads you own.
  • The subscription was used up, and the thread continued because the run-out setting is Keep going or someone clicked Use Capy credits.
  • Someone else owns the thread, so it runs on their connections even when you write in it. A thread that Slack or an automation started as someone else is theirs. The owner can use Transfer ownership in the thread's share dialog.
  • An admin turned off Members can connect their own subscriptions and keys.

Admins can exclude a company's models under Capy credits on the Models page. Never use Capy credits excludes every model. A thread then stops when no subscription or key can serve its model.

Subagent models⁠

Subagents use the parent thread's model by default. The agent can give a subagent a cheaper model for mechanical work or a stronger one for a hard subsystem.

When your balance runs out⁠

Your organization is out of credits means the balance can't pay for the thread's next step, either model usage on Capy credits or machine time. Your subscription's limits don't cause this message.

An admin can add balance, turn on auto-reload, or pick a plan from the card. A member sees Ask an organization admin to add balance. Then send a message or retry, because the thread doesn't resume on its own. You can also run the thread on your own computer.

An organization with no plan and no balance sees the Put Capy to work plan picker when it messages a cloud machine. A thread on your own computer that uses a subscription or key skips the plan picker.

Machine pricing⁠

Each thread's machine bills hourly by size, from $0.10 for Small to $3.20 for Big Guy. Large, the default, costs $0.40 an hour. You pay while a machine is awake, plus two minutes as it goes to sleep. A sleeping machine costs nothing. See machine sizes and pricing.

Enterprise⁠

For more than Capy Max's $1,000 a month in credits, custom pricing, or invoiced billing with net terms, contact us about a custom plan.

Open in Capy