NEWClaude Fable 5.1 is live on FableAI API

Build with Claude. Pay only for what you use.

Add credit with a coupon, create your own API keys and pay per token for every Claude model. No subscriptions, no surprises.

Coupon-based creditEvery Claude modelWorks with Claude Code
~/my-app — zshlive
claude-sonnet-5-5 · 1,284 in / 612 out$0.0087
How it works

From zero to first token in under a minute.

01
Add credit

Enter your coupon code on the Balance page — the credit lands instantly.

balance: $100.00
02
Create a key

Name it, cap its spend, pick allowed models. Shown once, stored hashed.

sk-ant-fai-7Hq2••••x9Lk
03
Start building

Point Claude Code or any Anthropic SDK at our base URL. Usage streams into your dashboard live.

ANTHROPIC_BASE_URL=…
Models

Every Claude model.
One wallet.

Discounted1M ctx
Fable 5.1

The flagship. Deep reasoning, long-horizon agents and the hardest code.

AgentsResearch
Per 1M tokens
$13.33
$20.00
−33%
Frontier1M ctx
Opus 5.5

Exceptional writing and analysis for complex, high-stakes work.

AnalysisWriting
Per 1M tokens
$12.50
Balanced1M ctx
Sonnet 5.5

The everyday workhorse. Powers Claude Code for most developers.

CodingProduction
Per 1M tokens
$12.90
Fastest200K ctx
Haiku 5.5

Near-instant responses for chat, routing and high-volume pipelines.

RealtimeScale
Per 1M tokens
$13.33
Cost calculator

Know the price
before you ship.

One simple price per model, per million tokens — input, output and cache tokens all count the same. Every request is metered at the upstream and deducted from your wallet to the fraction of a cent.

ModelPrice / 1M tokensContext
Fable 5.1newDiscounted
claude-fable-5-1
$13.33$20.001M
Opus 5.5newFrontier
claude-opus-5-5
$12.501M
Sonnet 5.5newBalanced
claude-sonnet-5-5
$12.901M
Haiku 5.5newFastest
claude-haiku-5-5
$13.33200K
Fable 5Discounted
claude-fable-5
$12.12$18.181M
Opus 5
claude-opus-5
$11.431M
Sonnet 5
claude-sonnet-5
$11.761M
Opus 4.8
claude-opus-4-8
$11.111M
Opus 4.7
claude-opus-4-7
$11.111M
Opus 4.6
claude-opus-4-6
$11.111M
Sonnet 4.6
claude-sonnet-4-6
$11.111M
Opus 4.5
claude-opus-4-5
$11.11200K
Sonnet 4.5
claude-sonnet-4-5-20250929
$11.11200K
Haiku 4.5
claude-haiku-4-5-20251001
$11.11200K
Opus 4.1
claude-opus-4-1-20250805
$11.11200K
Model
Input tokens / request4,000
Output tokens / request1,200
Requests / day300
Per request
$0.07
Per month
$628.49
Fable 5.1$668.40/mo
Opus 5.5$626.06/mo
Sonnet 5.5$628.49/mo
Haiku 5.5$741.00/mo
Drop-in compatible

Two lines. You're in.

# ~/.zshrc
export ANTHROPIC_BASE_URL="https://api.fableaiapi.com"
export ANTHROPIC_AUTH_TOKEN="sk-ant-fai-your-key"

claude   # that's it
Built for control

Everything a serious
team expects.

01 / LIMITS
Per-key spend limits

Cap each key in USD. Hit the cap and that key pauses while the rest keep running.

02 / LOGS
Real-time usage logs

Every request with billed tokens, latency and exact cost, exportable to CSV.

03 / CREDIT
Instant coupon credit

Enter a coupon code and the credit is on your balance in a second.

04 / KEYS
Unlimited keys

One per project, teammate or environment. Rename, disable or rotate any time.

05 / STATUS
Live status page

Health checks on every channel with latency and availability history, public at /status.

06 / KEYS
Per-key speed

Every key has its own requests-per-minute limit, so one busy app never slows the others.

FAQ

Questions,
answered.

View live system status

You add credit to your balance with a coupon code. Each request is metered and charged at the model's per-million-token price — every token (input, output, cache) counts the same. Your dashboard shows the exact billed tokens for every call. When your balance runs out, requests return a clear insufficient-balance error, never a surprise bill.

Get a coupon code from your admin and enter it on the Balance page. The credit is added instantly.

Yes. Set ANTHROPIC_BASE_URL to our endpoint and ANTHROPIC_AUTH_TOKEN to your key, then run claude. The official Anthropic SDKs work the same way — just change the base URL.

Yes. Every key can have its own USD spend cap, expiry date and model allow-list. When a key hits its cap it pauses while your other keys keep running.

No. We log request metadata for billing (model, token counts, latency, cost) — not prompt or response content.

Start building
in a minute.

Add credit with a coupon and spend it on any Claude model.