Rates · shared chat savings · media pricing · no subscription
Nyquest compresses context before it's billed. Whatever that saves against the provider's rate is the diff — half is credited straight back to your wallet, half is Nyquest's fee for eligible managed chat. Media is billed at provider cost. Sign up free with BYOK or any currently-free model; fund the wallet once to unlock Pro and every model in the catalog.
How the split works
Each eligible managed chat request is charged at the provider's rate for the model that served it — we don't mark up the base. In the same transaction, your half of what compression saved is credited back. The credit is subject to the provider-cost floor. Media generation uses the separate pricing below. There is no subscription.
| Line | What it is |
|---|---|
| Provider rate for the call | 100% |
| Tokens compression removes | the diff |
| Your half of the diff | credited back to your wallet |
| Nyquest's half | our managed-chat fee |
For illustration: a request that has a $1.00 input-token component before compression and no output charge in this simplified example. Removing two-fifths of those input tokens creates a $0.40 diff. $0.20 is credited back to you and $0.20 is Nyquest's fee, so the request nets $0.80. A request compression can't shrink has a $0.00 diff: you pay the provider's rate, and Nyquest earns nothing. Every credit is listed per request on the Savings page in the app. What compression saves on real traffic is measured on the blog.
What you pay. Eligible managed chat uses the model's catalog rates with a 50/50 split of compression savings, subject to the provider-cost floor. Media generation is billed at provider cost and is outside that split. BYOK is billed by your provider; free chat routes have no Nyquest usage charge. Wallet funding includes the processing fee shown at Checkout. Paid operations still need funds after Pro is activated.
Plans
Free gets you BYOK — your own provider keys, billed by the provider, outside the split — plus whichever models happen to be free in the catalog. Pro is one $5 fund, permanent: every model unlocks, limits go up, and the 50/50 split applies to eligible managed chat requests. Enterprise is the same split at your volume.
How Pro works: there is no subscription. Sign up free — you get BYOK plus access to whichever models are currently no-cost. The first time you fund your wallet — minimum $5 — your account flips to Pro permanently and every model in the catalog unlocks. From then on each eligible managed chat request is charged at the catalog rate and, in the same transaction, credited your half of the diff. No recurring charge, no auto-renewal, no expiring badge — when the balance runs down you're still Pro, you just have an empty wallet.
What costs what
Platform-hosted chat is billed from your prepaid wallet at the provider's rate — the base is never marked up — and compression credits half the saving back on every request. Rates track the upstream market and update when providers change theirs; the model picker always shows your live rate before you send.
| Tier | Examples | Input / 1M tokens | Output / 1M tokens |
|---|---|---|---|
| Cheap | Gemini 3.8 Flash (batch) · Gemini 3.8 Flash · Gemini 3.7 Flash | ~$0.15 – $1 | ~$0.60 – $5 |
| Mid | Claude Opus 5 (batch) · Grok 4.6 · Grok 4.5 | ~$2 – $3 | ~$10 – $15 |
| Top | Claude Fable 5.1 (batch) · Claude Fable 5.1 · GPT-6 Astra | ~$5 – $15 | ~$25 – $60 |
| Reasoning | o-series, DeepSeek R1 | ~$0.70 – $15 | ~$2.50 – $60 |
Typically $0.01 – $0.10 per image, varies by model.
A few cents per minute of audio.
Planner calls consume tokens at the served model's rates. Cost depends on context and steps; review the Agent Mode cost guide before running a goal.
Completely free on Nyquest and outside the split. Your provider charges you directly.
Detailed rates
Per-model reference rates below — the base your wallet is charged before your half of the diff comes back. The in-app picker is the source of truth for exactly what you'll pay. BYOK pays the provider directly. Refreshed daily from the live catalog (2026-09-05).
| Model | Provider | Input ($/1M) | Output ($/1M) |
|---|---|---|---|
| Anthropic | |||
| Claude Fable 5.1 (batch) | Anthropic | $5.00 | $25.00 |
| Claude Fable 5.1 | Anthropic | $10.00 | $50.00 |
| Claude Opus 5 (batch) | Anthropic | $2.50 | $12.50 |
| OpenAI | |||
| GPT-6 Astra | OpenAI | $10.00 | $50.00 |
| GPT-6 Astra (batch) | OpenAI | $5.00 | $25.00 |
| GPT-6 Astra Pro | OpenAI | $10.00 | $50.00 |
| Gemini 3.8 Flash (batch) | $0.38 | $1.88 | |
| Gemini 3.8 Flash | $0.75 | $3.75 | |
| Gemini 3.7 Flash | $0.75 | $3.75 | |
| xAI | |||
| Grok 4.6 | xAI | $2.00 | $6.00 |
| Grok 4.5 | xAI | $2.00 | $6.00 |
| Grok Build 0.1 | xAI | $1.00 | $2.00 |
| DeepSeek | |||
| DeepSeek V4 Flash Vision Exp | DeepSeek | $0.22 | $0.66 |
| DeepSeek V4 Pro 0813 (batch) | DeepSeek | $1.32 | $3.96 |
| Meta | |||
| Llama Guard 4 12B | Meta | $0.18 | $0.18 |
| Llama 4 Maverick | Meta | $0.20 | $0.70 |
| Mistral | |||
| Mistral Medium 3.5 (batch) | Mistral | $0.75 | $3.75 |
| Mistral Medium 3.5 | Mistral | $1.50 | $7.50 |
| Alibaba | |||
| Qwen3.8 Max (0902) | Alibaba | $2.00 | $6.00 |
| Qwen3.8 Flash | Alibaba | $0.15 | $0.47 |
418 models accessible in the picker. Rates refresh daily from the live catalog — provider changes flow through automatically. The model picker in-app always shows the live current rate before you send; your half of the diff is credited after.
Funding the wallet
Add credit via Stripe Checkout from the wallet balance control. Funds appear in your wallet within seconds. Card processing fees pass through to you — Stripe charges 2.9% + $0.30 per US-card transaction (international cards may be slightly higher), and that fee is added on top of the amount you choose to fund. Your wallet is credited the full amount you picked.
| You credit your wallet | Approx. Stripe fee | Total card charge |
|---|---|---|
| $5.00 | ~$0.46 | ~$5.46 |
| $10.00 | ~$0.61 | ~$10.61 |
| $25.00 | ~$1.06 | ~$26.06 |
| $50.00 | ~$1.81 | ~$51.81 |
| $100.00 | ~$3.30 | ~$103.30 |
| $500.00 | ~$15.25 | ~$515.25 |
Minimum funding $5. Maximum per top-up $500. Wallet balance never expires and is refundable for unspent amounts within ~7 days of funding. See refund policy →
What's always free
Regardless of whether you've funded a wallet, these are always available at no cost.
Creation, sign-in, OAuth, password reset.
Browsing past conversations, artifacts, and projects.
Help center and documentation.
Switching between the 12 chassis themes.
Your provider, your bill — outside the split.
The layer is free; chat that builds memories pays the chat rate.
Settings, account management, billing, and the Savings page.
Built in — no separate Tavily/SerpAPI key.
Common questions
No. Pro is a one-time activation triggered by your first wallet funding. After that, you're Pro forever — no recurring charge, no auto-renewal. Eligible managed chat shares compression savings; media generation is billed at provider cost.
The tokens compression removed from a request before it was sent, priced at the provider's rate for the model that served it. It is measured per request, when the request is charged — not estimated at month end.
As a credit to your wallet in the same transaction as the request's charge. The Savings page in the app keeps the ledger — tokens saved for the period, the split by model, and each recent request as prompt → sent tokens with the credit for it.
Then the diff is zero: you pay the provider's rate, nothing more, and Nyquest earns nothing on that request. Short, history-free questions are usually like this. Long conversations are where the diff grows.
One floor: a request never costs you less than what the provider charged Nyquest for it. That can happen when a request carries large memory or web-grounding context you weren't billed for. On those, the credit is held at the floor, and the Savings page marks them.
The split is the same at every volume — nobody gets a better rate for being big or a worse one for being small. Enterprise adds custom agreements, limits, and support rather than a different split. At $1,000+/month, reach out via live support.
Not yet. The wallet doesn't auto-refill — you fund manually via the wallet balance control. A "balance below $X → auto-charge $Y" toggle is on the roadmap.
Platform-hosted requests return a clear "wallet empty, please fund" error. BYOK requests continue to work because they don't touch the wallet. Nothing is interrupted permanently — fund whenever and you're back.
No. Most chat and media charges settle after provider work completes; video reserves an upfront hold. In-flight or concurrent usage can settle after the displayed balance changes. Check Usage and existing jobs before repeating a request near zero balance. Wallet details.
Unspent funds are refundable on request, generally within 7 days of funding. See Refunds and Disputes for the full policy.
Every Stripe charge generates a Stripe receipt to your email. For formal invoices (business / tax purposes), open live support and we'll send a PDF.
We don't currently collect VAT or sales tax for individual users in most jurisdictions. Stripe Checkout shows whatever applies in your region. Business customers needing tax-compliant invoices should contact support.
Pin if you want a fixed base rate — every message in that conversation uses that model at its published rate, and the split still applies on top. Auto-route lets the dispatcher pick by tier and complexity; cost varies request-to-request, and the actual model, cost, and credit are always shown per request.
No card required. BYOK and any currently-free models cost nothing. One $5 fund unlocks Pro, every model, and the 50/50 split on every request.
Open NyquestSplit the Diff · BYOK always free · No subscriptions · Live per-request pricing