Transparent value · No hidden fees

Plans that grow with you

You are charged exactly the price shown. Each plan includes premium and Hanzo model usage each month.

Save 17% with annual

Free

Meet Hanzo. Chat and build with Hanzo's free models.

Free

No card needed

Try Hanzo

No card needed · Instant access

Hanzo's free models: enso-free and zen-free
Chat on free models, with limited usage
The app's core features
Upgrade any time from inside the app

Pro

For one person: every room in the hanzo.ai app, with premium and Hanzo model usage included each month.

$20 USD/month

Billed monthly

Try Hanzo

No commitment · Cancel anytime

Everything in Free and:
Every room in the hanzo.ai app
Includes premium and Hanzo model usage each month
Top up prepaid credit anytime for more
Limited free usage when you reach your plan's included usage
MCP, the SDK and the CLI included

Max

More included usage than Pro, with Enso orchestration, managed agents and a resident bot.

From $100 USD/month

Billed monthly

Try Hanzo

No commitment · Cancel anytime

Everything in Pro, plus:
More included usage than Pro
Includes premium and Hanzo model usage each month
Top up prepaid credit anytime for more
Limited free usage when you reach your plan's included usage
Enso orchestration — dedicated routing through Hanzo's foundational framework model
Managed agents and a resident bot, scheduled in the background
Priority GPU inference queues

Each plan includes premium and Hanzo model usage each month. When it runs out you keep chatting with limited free usage, or choose to continue with prepaid credit, which you can top up anytime.

What a bill is made of

Four components, each on your spend breakdown under its own name. Two are quoted before they run; two are metered from what they used.

Model inference

per token, five tiers

Quoted before it runs

Input, cached reads, cache writes, output and reasoning are metered as five disjoint tiers, at the tariff for the completion window the request asked for.

inputcache_readcache_writeoutputreasoning

Computer

per second, while running

Quoted before it runs

vCPU and memory on what the machine was given. Nothing to create one, nothing while it sleeps, and the boot disk is inside the allowance — so a computer waiting between turns meters nothing.

Web tools

per call

Metered from what it used

Priced from what the provider charged, so a web call is booked after it runs. A fetch the domain policy refuses is stopped before the request and costs nothing.

Media generation

what the job cost

Metered from what it used

Billed once per finished job at its real cost; a render cannot be quoted before it runs, so it is admitted against the headroom under the agent's per-task cap instead. A job that fails is not billed.

The rates

Every rate here is the price you pay — the number the ledger books and the number your spend breakdown reports.

	Rate
Computer · vCPU	$0.0504/hr$0.000014/second	Charged by the second, while running.
Computer · memory	$0.0162/GiB-hr$0.0000045/GiB-second	Charged by the second, on the memory the machine was given.
Computer · paused	$0	No vCPU, no memory, no storage.
Computer · created	$0	There is no creation fee.
web_search	$0.002158/call	One query, up to ten results.
web_fetch	$0.001079/call	One page read as text.
MCP, SDK and CLI	$0	The interfaces are not metered.
Seats	$0	No per-seat fee, however many people use the organization.

The completion window is the biggest lever

The same model at a different window is a different price, chosen per request. Pay the interactive tariff only when a person is waiting.

immediateAnswers now, at the highest tariff.

priorityAnswers soon, at a lower one.

looseAnswers eventually, at the lowest.

Every model, at the rate the ledger books

Model inference is metered per token in five disjoint tiers, at the tariff for the completion window the request asked for.

14Zen Models

397Third-Party

436Total Models

57Providers

11Featured

74Free Models

Hanzo Enso

Flagship frontier models — Flash, Pro, and Ultra. Live per-token retail, one API key.

Explore Enso

Enso Flash

pro

Fast, economical Enso tier for high-volume, low-latency everyday work, with 1M-context overflow.

1M context window

Low latency

Start Using enso-flash

Input$2 / MTok

Output$4 / MTok

Enso

ultra max

Hanzo's proprietary frontier model — Opus-class reasoning by default with 1M-context overflow.

1M context window

Frontier reasoning

Start Using enso

Input$4 / MTok

Output$20 / MTok

Enso Ultra

ultra max

Adaptive fan-out — probes a task-appropriate model, escalates to a top-K panel only when needed, then verifies-then-selects the best answer.

200K context window

Adaptive fan-out

Start Using enso-ultra

Input$5 / MTok

Output$25 / MTok

Zen Models

32 models across LLM, embedding, reranker, image, audio, and videoUpdated Aug 6, 2026

Full Model Catalog

ultra max

ultra

pro max

pro

Zen Guard — Content Safety

starter

Safety classifier for moderation and guardrails across a broad category and language set.

128K context

Safety classifier

Input$0.3 / MTok

Output$0.96 / MTok

Cache ReadFree

Zen VL — Vision-Language

pro

Reads images and reasons over them — visual Q&A, document and chart understanding, grounded captioning.

128K context

Vision + Language

Start Using zen-vl

Input$0.09 / MTok

Output$0.39 / MTok

Cache ReadFree

Zen5 (Default)

pro

Canonical Zen5 default. a 35B frontier MoE (3B active) (35B total / 3B active per token, released Apr 2026, Apache-2.0). 256K context, agentic-trained, OpenAI + Anthropic API. The everyday Zen5 chat model.

Parameters: 35B (3B active)Architecture: Zen frontier MoE

256K context window

35B total / 3B active (MoE)

Zen base

Apr 2026 release

OpenAI + Anthropic API

Start Using zen5

Input$2.28 / MTok

Output$7.26 / MTok

Cache Read$0.42 / MTok

Zen5 Coder

pro

Code-specialized Zen5 tier. 80B sparse MoE tuned for repo-scale code understanding, agentic refactoring, and tool-use coding loops.

Parameters: 80B (MoE)Architecture: Zen Coder MoE

256K context

80B sparse MoE

Code-specialized

Agentic / tool-use

Start Using zen5-coder

Input$2.28 / MTok

Output$7.26 / MTok

Cache Read$0.42 / MTok

Zen5 Flash

starter

Smallest and cheapest text-only Zen5 chat tier. 4B-class dense, sub-100ms TTFT, 32K context. For high-volume routing and simple agent loops.

Parameters: 4BArchitecture: Zen dense

32K context window

4B parameters (dense)

Sub-100ms TTFT

Highest throughput

Start Using zen5-flash

Input$0.2646 / MTok

Output$0.5292 / MTok

Cache Read$0.05292 / MTok

Zen5 Mini

pro

Frontier agentic at the lowest cost in the family. Built on a 230B agentic MoE (10B active) (230B MoE / 10B active, released Feb 2026). 80.2% SWE-Bench Verified, 76.3% BrowseComp; trained on 200K+ real-world environments via large-scale RL.

Parameters: 230B (10B active)Architecture: Zen MoE

192K context window

230B total / 10B active (MoE)

Zen agentic base

Frontier agentic / coding

Lowest $/token in family

Start Using zen5-mini

Input$0.09 / MTok

Output$0.39 / MTok

Cache Read$0.09 / MTok

Zen5 Pro

ultra

Zen Flash IQ2_XXS-imatrix weights (81 GB GGUF on zenlm/zen-5-pro-gguf). 284B total / 37B active per token, 1M context, asymmetric routed-MoE quant. Fits a single 128 GB Apple Silicon / DGX Spark / H100 80 GB.

Parameters: 284B (37B active)Architecture: Zen Flash MoE

1M context window

284B total / 37B active (MoE)

Zen Flash base

IQ2_XXS-imatrix quant (81 GB)

Runs on 128 GB hardware

Start Using zen5-pro

Input$1.305 / MTok

Output$2.61 / MTok

Cache Read$0.010875 / MTok

Router General

pro

Automatic model routing — hosted by Hanzo.

Architecture: router

Automatic model routing

InputN/A

OutputN/A

Router Knowledge Base Document

pro

Automatic model routing — hosted by Hanzo.

Architecture: router

Automatic model routing

InputN/A

OutputN/A

Router Software Engineering

pro

Automatic model routing — hosted by Hanzo.

Architecture: router

Automatic model routing

InputN/A

OutputN/A

Router Software Engineering 01

pro

Automatic model routing — hosted by Hanzo.

Architecture: router

Automatic model routing

InputN/A

OutputN/A

Router Writing

pro

Automatic model routing — hosted by Hanzo.

Architecture: router

Automatic model routing

InputN/A

OutputN/A

All Zen LLMs available via OpenAI-compatible API at api.hanzo.ai/v1/chat/completions

Embedding Models

High-quality text embeddings via /v1/embeddings

Zen Embedding

starter

Dense text embeddings for semantic search, retrieval, and clustering.

8K context

Embeddings

Input$0.06 / MTok

Output$0.06 / MTok

Cache ReadFree

All Mini Lm L6 V2

pro

Text embeddings — hosted by Hanzo.

Architecture: embedding

Text embeddings

Input$0.02 / MTok

OutputFree

Bge M3

pro

Text embeddings — hosted by Hanzo.

Architecture: embedding

Text embeddings

Input$0.02 / MTok

OutputFree

E5 Large V2

pro

Text embeddings — hosted by Hanzo.

Architecture: embedding

Text embeddings

Input$0.02 / MTok

OutputFree

Gte Large En V1.5

pro

Text embeddings — hosted by Hanzo.

Architecture: embedding

Text embeddings

Input$0.02 / MTok

OutputFree

Multi Qa Mpnet Base Dot V1

pro

Text embeddings — hosted by Hanzo.

Architecture: embedding

Text embeddings

Input$0.02 / MTok

OutputFree

Qwen3 Embedding 0.6b

pro

Text embeddings — hosted by Hanzo.

Architecture: embedding

Text embeddings

Input$0.02 / MTok

OutputFree

Reranker Models

Improve retrieval quality via /v1/rerank

Zen Rerank

starter

Cross-encoder reranker — scores (query, document) pairs to reorder retrieval results.

Reranking

Retrieval

Price$0.03

Bge Reranker V2 M3

pro

Reranking — hosted by Hanzo.

Architecture: rerank

Reranking

Input$0.02 / MTok

OutputFree

Image Generation

FLUX, Stable Diffusion, and more via /v1/images/generations

Zen Image

pro

Text-to-image generation.

Text → Image

Price$0.24 / image

Openai Gpt Image 1

pro

Text-to-image — hosted by Hanzo.

Architecture: image

Text-to-image

Price$0.04 / image

/v1/images/generations

Openai Gpt Image 1.5

pro

Text-to-image — hosted by Hanzo.

Architecture: image

Text-to-image

Price$0.04 / image

/v1/images/generations

Openai Gpt Image 2

pro

Text-to-image — hosted by Hanzo.

Architecture: image

Text-to-image

Price$0.04 / image

/v1/images/generations

Stable Diffusion 3.5 Large

pro

Text-to-image — hosted by Hanzo.

Architecture: image

Text-to-image

256 context window

Price$0.04 / image

/v1/images/generations

Audio & Speech

Speech-to-text, text-to-speech, and audio generation via /v1/audio/*

Zen Foley

pro

Text-to-sound-effects — generates Foley and ambient audio from a prompt.

Text → Sound FX

Price$0.15

Zen Music

pro

Text-to-music generation.

Text → Music

Price$0.3

Zen Voice

pro

Text-to-speech — natural voice synthesis.

Text → Speech

Price$0.045

Qwen3 Tts Voicedesign

pro

Text-to-speech — hosted by Hanzo.

Architecture: speech

Text-to-speech

33k context window

Price$5

/v1/audio/speech

Video Generation

Text-to-video via /v1/videos/generations

Zen Video

pro max

Text-to-video — generates short clips from a prompt (async).

Text → Video

Async

Price$1.8

Wan2 2 T2v A14b

pro

Text-to-video — hosted by Hanzo.

Architecture: video

Text-to-video

100 context window

Price$0.5

/v1/videos/generations

Customers can purchase prioritized API capacity with Priority Tier

Prompt caching pricing is for the standard 5-minute TTL. A longer one costs more.

Try Hanzo

Hanzo Kai

The decision model: a typed answer to each question, via /v1/decisions.

Explore Kai

Kai

starter

Hanzo's decision model. $0.021 per million input tokens, half of Jev's $0.042. Output is free.

Architecture: decision

Decisions

Calibrated probabilities

Input$0.021 / MTok

OutputFree

/v1/decisions

Explore pricing for tools

Web Search

per call$0.002158

Code Interpreter

per session minute$0.03

File Storage

per GB/month$0.2

Image Generation

per image$0.04

Speech-to-Text

per minute$0.006

Text-to-Speech

per 1M characters$15

Web Fetch

per call$0.001079

*Does not include input and output tokens required to process requests

Try Hanzo

Featured Third-Party Models

Top models from leading AI providers — all accessible through one Hanzo API key

Anthropic: Claude Opus 4.6

Anthropic

Featured

1M context windowtext+image+file->text

Input$5 / MTok

Output$25 / MTok

Anthropic: Claude Sonnet 4.6

Anthropic

Featured

1M context windowtext+image+file->text

Input$3 / MTok

Output$15 / MTok

Anthropic: Claude Haiku 4.5

Anthropic

Featured

200k context windowtext+image+file->text

Input$1 / MTok

Output$5 / MTok

OpenAI: GPT-5

OpenAI

Featured

400k context windowtext+image+file->text

Input$1.25 / MTok

Output$10 / MTok

OpenAI: GPT-5 Mini

OpenAI

Featured

400k context windowtext+image+file->text

Input$0.25 / MTok

Output$2 / MTok

Google: Gemini 2.5 Pro

Google

Featured

1M context windowtext+image+file+audio+video->text

Input$1.25 / MTok

Output$10 / MTok

DeepSeek: R1

DeepSeek

Featured

164k context windowtext->text

Input$0.7 / MTok

Output$2.5 / MTok

DeepSeek: DeepSeek V3

DeepSeek

Featured

164k context windowtext->text

Input$0.257 / MTok

Output$1.03 / MTok

Meta: Llama 4 Maverick

Mistral: Mistral Large 3 2512

Mistral

Featured

262k context windowtext+image+file->text

Input$0.5 / MTok

Output$1.5 / MTok

Cohere: Command A

Cohere

Featured

256k context windowtext->text

Input$2.5 / MTok

Output$10 / MTok

All Third-Party Models

397 models from 57 providers — dynamically detected and priced

397 models

Model	Provider	Context	Input	Output
Anthropic: Claude Opus 4.6 Featured	Anthropic	1000K	$5 / MTok	$25 / MTok
Anthropic: Claude Sonnet 4.6 Featured	Anthropic	1000K	$3 / MTok	$15 / MTok
Anthropic: Claude Haiku 4.5 Featured	Anthropic	200K	$1 / MTok	$5 / MTok
OpenAI: GPT-5 Featured	OpenAI	400K	$1.25 / MTok	$10 / MTok
OpenAI: GPT-5 Mini Featured	OpenAI	400K	$0.25 / MTok	$2 / MTok
Google: Gemini 2.5 Pro Featured	Google	1049K	$1.25 / MTok	$10 / MTok
DeepSeek: R1 Featured	DeepSeek	164K	$0.7 / MTok	$2.5 / MTok
DeepSeek: DeepSeek V3 Featured	DeepSeek	164K	$0.257 / MTok	$1.03 / MTok
Meta: Llama 4 Maverick Featured	Meta	1049K	$0.2 / MTok	$0.8 / MTok
Mistral: Mistral Large 3 2512 Featured	Mistral	262K	$0.5 / MTok	$1.5 / MTok
Cohere: Command A Featured	Cohere	256K	$2.5 / MTok	$10 / MTok
Gemma SEA LION v4 27B IT Free	AI Singapore	N/A	Free	Free
Qwen SEA LION v4 32B IT Free	AI Singapore	N/A	Free	Free
AI21: Jamba Large 1.7	AI21	256K	$2 / MTok	$8 / MTok
AionLabs: Aion-2.0	Aion Labs	131K	$0.8 / MTok	$1.6 / MTok
AionLabs: Aion-3.0	Aion Labs	131K	$3 / MTok	$6 / MTok
AionLabs: Aion-3.0-Mini	Aion Labs	131K	$0.7 / MTok	$1.4 / MTok
AionLabs: Aion-RP 1.0 (8B)	Aion Labs	33K	$0.8 / MTok	$1.6 / MTok
AllenAI: Olmo 3 32B Think	Allen AI	66K	$0.15 / MTok	$0.5 / MTok
Olmo 3 7B Instruct Free	Allen AI	N/A	Free	Free
Amazon: Nova 2 Lite	Amazon	1000K	$0.3 / MTok	$2.5 / MTok
Amazon: Nova Lite 1.0	Amazon	300K	$0.06 / MTok	$0.24 / MTok
Amazon: Nova Micro 1.0	Amazon	128K	$0.035 / MTok	$0.14 / MTok
Amazon: Nova Premier 1.0	Amazon	1000K	$2.5 / MTok	$12.5 / MTok
Amazon: Nova Pro 1.0	Amazon	300K	$0.8 / MTok	$3.2 / MTok
Magnum v4 72B	Anthracite	16K	$3 / MTok	$5 / MTok
Anthropic Claude Haiku Latest	Anthropic	200K	$1 / MTok	$5 / MTok
Anthropic Claude Sonnet Latest	Anthropic	1000K	$2 / MTok	$10 / MTok
Anthropic: Claude 3 Haiku	Anthropic	200K	$0.25 / MTok	$1.25 / MTok
Anthropic: Claude Fable 5	Anthropic	1000K	$10 / MTok	$50 / MTok
Anthropic: Claude Fable Latest	Anthropic	1000K	$10 / MTok	$50 / MTok
Anthropic: Claude Opus 4	Anthropic	200K	$15 / MTok	$75 / MTok
Anthropic: Claude Opus 4.1	Anthropic	200K	$15 / MTok	$75 / MTok
Anthropic: Claude Opus 4.5	Anthropic	200K	$5 / MTok	$25 / MTok
Anthropic: Claude Opus 4.7	Anthropic	1000K	$5 / MTok	$25 / MTok
Anthropic: Claude Opus 4.7 (Fast)	Anthropic	1000K	$30 / MTok	$150 / MTok
Anthropic: Claude Opus 4.8	Anthropic	1000K	$5 / MTok	$25 / MTok
Anthropic: Claude Opus 4.8 (Fast)	Anthropic	1000K	$10 / MTok	$50 / MTok
Anthropic: Claude Opus Latest	Anthropic	1000K	$5 / MTok	$25 / MTok
Anthropic: Claude Sonnet 4	Anthropic	1000K	$3 / MTok	$15 / MTok
Anthropic: Claude Sonnet 4.5	Anthropic	1000K	$3 / MTok	$15 / MTok
Anthropic: Claude Sonnet 5	Anthropic	1000K	$2 / MTok	$10 / MTok
Claude Opus 5	Anthropic	1000K	$5 / MTok	$25 / MTok
Claude Opus 5 (Fast)	Anthropic	1000K	$10 / MTok	$50 / MTok
Arcee AI: Trinity Large Thinking	Arcee	262K	$0.22 / MTok	$0.85 / MTok
Arcee AI: Virtuoso Large	Arcee	131K	$0.75 / MTok	$1.2 / MTok
Baidu: ERNIE 4.5 VL 424B A47B	Baidu	123K	$0.42 / MTok	$1.25 / MTok
ERNIE 4.5 VL 424B A47B Base PT Free	Baidu	N/A	Free	Free
ByteDance Seed: Seed 1.6	ByteDance	262K	$0.25 / MTok	$2 / MTok
ByteDance Seed: Seed 1.6 Flash	ByteDance	262K	$0.075 / MTok	$0.3 / MTok

Third-party model pricing includes a passthrough markup. Prices update automatically every 6 hours. All models accessible via api.hanzo.ai/v1/chat/completions using the provider-prefixed model ID (e.g. anthropic/claude-sonnet-4.6).

Frequently Asked Questions

What is free?

Running your own private AI on your own machine, and the open Zen weights, free forever. The free models on the API, on a daily pool. Everything open source is yours to download and run.

What does a plan cost?

Exactly the price on its card, by the month or, where a year is sold, by the year. Free costs nothing. Team is priced per seat, from 2 seats. Each plan includes premium and Hanzo model usage each month; top up prepaid credit anytime for more. Enterprise is priced with our team.

What happens when I reach my plan's included usage?

You keep chatting with limited free usage on free models. Turn on Continue with credits to keep using any model and pay from your prepaid credit, or upgrade for more included usage. Included usage renews each month.

What is a bill made of?

Four components, and every charge is one of them: model inference, the computer an agent runs on, web tools, and media generation. Each appears on your spend breakdown under its own name, so a month adds up to four lines you can check rather than one you have to trust.

Which charges are quoted before they run?

Model calls and compute. Both can be bounded from above before they happen, so an agent that would breach its budget is refused rather than discovered on an invoice. Web calls and media renders cannot be — a provider's charge and a render's cost are only known afterwards — so those are booked from what they actually cost, and a render is admitted against the headroom left under the agent's per-task cap.

How are tokens billed?

Per token, in five tiers that never overlap: input, cached reads, cache writes, output and reasoning, at the tariff for the completion window the request asked for. Amounts are carried as integer micro-dollars, so nothing rounds in our favor.

What is a completion window?

How soon you need the answer, chosen per request, and the biggest lever you have on what a call costs — the same model at a different window is a different price. Ask for it immediately and you pay the interactive tariff; ask for it soon, or eventually, and you pay less. Pay the interactive rate only when a person is actually waiting.

What does a computer cost when nothing is running on it?

Nothing. vCPU and memory meter by the second while the machine runs, on what it was given. Creating one is free, a paused one has no vCPU, no memory and no storage, and its boot disk is inside the allowance — so an agent waiting between turns meters nothing at all.

Are MCP, the SDK and the CLI metered?

No. The interfaces are free; only the work they ask for is metered. Connecting an editor, a terminal or an agent framework costs nothing by itself.

How are web tools billed?

Per call, priced from what the provider charged us, so a web call is booked after it runs rather than quoted before it. A fetch your own domain policy refuses never leaves the process, so it costs nothing.

How is media generation billed?

Once per finished job, at what the job actually cost. A render cannot be quoted before it runs, so it is admitted against the headroom under the agent's per-task cap instead of against a quote. A job that fails is not billed.

Are failed calls billed?

No. A call or a job that fails is not billed, and any hold it placed against a budget comes back. A request your own policy refuses is stopped before it is sent and costs nothing.

Do you mark up model prices?

The rate shown beside a model is the number the ledger books — there is no list price that differs from the billed one, and the rates on this page are read from the same place the meter reads them. Our own costs are measured against vendor invoices rather than published rate cards, which overstate real cost substantially.

Is streaming billed differently?

No. The same tokens at the same tariff, streamed or not.

How do budgets work?

Every agent carries one: a cap per period and a cap per task. A call that can be quoted is quoted before it runs, and if it would breach the cap it is refused rather than discovered on an invoice. The agent is told in-band, so it can wrap up rather than crash, and a session paused at its cap resumes when the cap is raised. The exact fields and the order of the checks are in the docs.

What are the rate limits?

The free pool is counted per day. Paid accounts get high limits; Enterprise can arrange dedicated ones.

Can I separate environments?

Yes. Keys belong to an organization and can be scoped to a project or a service account, so development, staging and production carry their own keys and their own spend.

How do I migrate from OpenAI or Anthropic?

Point your client at api.hanzo.ai/v1 with a Hanzo key. The chat, embeddings and tool-calling surfaces are compatible, and the same models answer at the same endpoint alongside every other lab’s. Tools work natively and over MCP at api.hanzo.ai/v1/mcp, which is how Claude Code, Cursor, Codex and the rest connect.

What payment methods do you accept?

All major cards, and crypto on pay-as-you-go. Enterprise customers can arrange invoicing and PO terms. Payments are processed by our payment provider; we never store your card.

Is there a minimum spend or a lock-in?

No minimum. A plan renews for the term you bought, a month or a year. Cancel any time; what you already paid for stays yours until the period ends.

Do you offer annual billing?

Yes. A plan sold by the year can be bought for a whole year at once, which saves up to 20% against paying monthly. Turn on the annual switch under the plans to see each yearly price.

Do you train on my data?

No. Your content is yours. We use industry-standard encryption in transit and at rest, and you can request deletion at any time.

Do you support SSO?

Enterprise accounts sign in through their own identity provider over OIDC or SAML, with SCIM provisioning and org-scoped roles.

Where do I check uptime and incidents?

status.hanzo.ai, which every dashboard links to.

Do you offer pricing for education or nonprofits?

Yes, for both. Contact sales with your institution or nonprofit credentials and we will sort out the right rate.

Measured, not modelled.

Every figure is corrected against vendor invoice brackets — list-price rate cards overstate real cost by up to 4.2x. The method, and the cost per completed task, are on the bench.

What the same stack costs to run yourself: the comparison and the calculator. Starting a company on it: the startup programme.