English
Sign in
Claude · GPT · DeepSeek · GLM · Kimi & more

Admesh.ai_
Every major model. Always on. Lower prices.

Claude, GPT, DeepSeek, GLM, Kimi and more through one gateway — automatic multi-cloud failover keeps you online, and pay-as-you-go pricing costs less than going direct.

Start free · claim credit See pricing
$curl https://api.admesh.ai/v1/chat/completions -H "Authorization: Bearer $ADMESH_API_KEY"
Live gateway · api.admesh.ai/v1
Live
anthropic/claude-fable-5Claude Fable 5 · 1M
$3.25/1M
google/gemini-2.5-flashGemini 2.5 Flash · 1.0M
$0.135/1M
moonshot/kimi-k2.5Kimi K2.5 · 262K
$0.325/1M
Gateway statusLive

One protocol

OpenAI-compatible. Swap the Base URL and you're done — zero refactoring.

Multi-channel failover

On 429 or outage, switch to a backup channel for the same model.

Pay-as-you-go, transparent

No hidden fees. Metered to the cent, with a log for every request.

01. Transparent · Live pricing

Models & live rates

Official-equivalent pricing, metered per token, with cache discounts

Model & IDContext windowInput / 1MOutput / 1MCached input / 1MStatus
Claude Fable 5
anthropic/claude-fable-5
1M$3.25 / 1M$16.25 / 1M$0.325 / 1MReady
Gemini 2.5 Flash
google/gemini-2.5-flash
1.0M$0.135 / 1M$1.125 / 1M$0.0135 / 1MReady
Kimi K2.5
moonshot/kimi-k2.5
262K$0.325 / 1M$1.625 / 1M$0.0975 / 1MReady
GPT-5.5
openai/gpt-5.5
1.1M$0.3 / 1M$1.8 / 1M$0.03 / 1MReady
GLM-4.5-Air
zhipu/glm-4.5-air
131K$0.0715 / 1M$0.4675 / 1M$0.01375 / 1MReady
Claude Fable 5.1
anthropic/claude-fable-5-1
1M$3.25 / 1M$16.25 / 1M$0.08125 / 1MReady
Claude Haiku 4.5
anthropic/claude-haiku-4-5-20251001
200K$0.325 / 1M$1.625 / 1M$0.0325 / 1MReady
Claude Haiku 5.5
anthropic/claude-haiku-5-5
1M$0.0325 / 1M$0.1625 / 1M$0.00325 / 1MReady
* Standard public-channel rates; high-concurrency channels get tiered discountsSee all 49 models and live rates
02. Plans & pricing

Pure pay-as-you-go. No monthly fee.

No rigid subscriptions. Top up when you need, credit never expires, and one key routes every major model.

No sunk cost · Top up anytime · Credit never expires · Billed only for usage
No monthly feeTry & validate

Developer Starter

For individual devs, scripts, and trying new models

$20/ from $50
Pay as you go — full balance becomes token credit
What's included
Credit never expires or resets
Every model: Claude, GPT, DeepSeek & more
Any OpenAI-compatible client or lightweight agent
Precise per-request billing details
Start with credit
Most popular
Most popularHigh-throughput

Team Pay-as-you-go

For production and ongoing builds

$100/ from $500
Self-serve top-up, priority routing
What's included
Full feature set, credit never expires
Multi-key quotas and group permissions
High concurrency with automatic failover
WeChat / Alipay / card billing + invoices
Top up & go
EnterpriseCustom & dedicated

Enterprise Plan

A compliant, elastic model hub for enterprise teams

Custom
Framework agreements & postpaid credit lines
What's included
Dedicated clusters with high QPS limits
Invoice settlement and VAT receipts
SLA and 7×24 dedicated support
Audit log export and usage dashboards
Contact sales
Enterprise & private deployment

Unified team AI access, credit and cost management

Dedicated API access, member groups with key-level permissions, per-member and per-model cost analytics, contract invoicing and corporate settlement.

Talk to sales
03. Reliable · Financial-grade

Why choose Admesh.ai

More than a proxy — an enterprise routing hub with high availability, resilience and cost governance

One protocol, zero migration

Strict OpenAI-compatible API. LangChain, Cursor, NextChat or your own app — swap the domain and key, no extra dependencies.

POST /v1/chat/completionsAuthorization: Bearer sk-live-••••model: anthropic/claude-fable-5

Multi-cloud, millisecond failover

Built-in adaptive probes. On 502/503, 429, or latency spikes, the gateway hot-switches to a backup channel with zero downtime.

Primary · direct→Backup A→Backup BAuto switch

Financial-grade billing

No black-box pricing. Every request has a unique Trace ID, token breakdown, and timestamp. Export reports and see exactly where budget goes.

request_idreq_8f2a…c91d
input_tokens1,284
output_tokens512
charge$0.0132
04. FAQ

Questions? We have answers

Integration, models, failover, and billing

Admesh.ai is a high-availability model gateway. We wrap Claude, GPT, DeepSeek, GLM, Kimi and more into one OpenAI-compatible API, with load balancing, millisecond failover, and usage management — so you stop juggling provider accounts.
Your credit never expires. Admesh.ai is 100% pay-as-you-go with no subscriptions or forced monthly fees, and your balance works across every model on the platform.
Set your Base URL to https://api.admesh.ai/v1 and paste an API key from the console. Existing OpenAI Python / Node SDKs work without changing your call logic.
A single model can be backed by multiple upstream channels. On timeout, rate limiting (429), or 5xx errors the gateway switches to a backup channel for the same model; repeated failures trip a circuit breaker until health probes recover. Switching only happens within the same model — we never silently swap Claude for a different model.
WeChat Pay, Alipay, card, and corporate wire transfer. After spending, you can request a digital invoice from the console under Billing — delivered to your email within 1–2 business days.