All migrations
Migration guide

From AIMLAPI to ClaudeAPI.cheap.
Trust signals you can verify.

Cheap is easy to claim. The hard part is doing it without silent quantization, a black-hole support inbox, or vague refund policies. We publish a status page anyone can hit, refund within 7 days, and never substitute lower-precision weights. Drop in via the same OpenAI SDK — and verify every claim from the URL bar.

99.5% uptime
Pay with crypto
Balance never expires
Named human support
No quantization

Where the trust signals diverge

Every row below is a yes/no question with an external URL you can check yourself.

Trust signal
AIMLAPI
ClaudeAPI.cheap
Public uptime page
Not surfaced
status.claudeapi.cheap (1-min ping)
SLA with credit clause
Not posted
99.5% rolling 90d, 1% credit if breached
Refund policy
Case-by-case
Full refund within 7 days, public terms
Named human support
Chatbot first
support@claudeapi.cheap, 24h reply business days
Quantization disclosure
Silent / unclear
No quantization, weekly parity benchmark
Payment method
Card on file
Crypto prepaid only, no auto-charge surprise
Balance expiry
Implicit
Never expires, ever

Read the full SLA terms on /trust. The status page link works without login.

Pricing is competitive — but it's not the lead

Per 1M tokens, USD. Anthropic list prices shown for reference.

Model
Anthropic list
ClaudeAPI.cheap Pro
Off list
Claude Opus 4.7
in / out per 1M tokens
$5.00 / $25.00
$1.00 / $5.00
−80%
Claude Sonnet 4.6
in / out per 1M tokens
$3.00 / $15.00
$0.60 / $3.00
−80%
Claude Haiku 4.5
in / out per 1M tokens
$1.00 / $5.00
$0.20 / $1.00
−80%

See /claude-api-pricing-guide for full per-model rates including GPT-5 and Gemini 3.

Migration in 3 steps

1

Get an API key (30 seconds)

Sign up at claudeapi.cheap/signup — email + crypto, no card on file. Generate an sk-cc-... key.

2

Swap base URL + key

AIMLAPI uses the OpenAI SDK shape. So do we. Two lines change:

Before (AIMLAPI)
python
from openai import OpenAI client = OpenAI( base_url="https://api.aimlapi.com/v1", api_key="aimlapi-your-key", ) resp = client.chat.completions.create( model="claude-3-opus-20240229", messages=[{"role": "user", "content": "Hello"}], )
After (ClaudeAPI.cheap)
python
from openai import OpenAI client = OpenAI( base_url="https://claudeapi.cheap/api/proxy/v1", api_key="sk-cc-your-claudeapi-cheap-key", ) resp = client.chat.completions.create( model="claude-opus-4-7", messages=[{"role": "user", "content": "Hello"}], )

Native Anthropic SDK works too — use base_url="https://claudeapi.cheap/api/proxy".

3

Verify parity against your old AIMLAPI baseline

Run a fixed prompt against AIMLAPI and against us. Compare outputs side by side. Our response should match Anthropic Direct exactly (no quantization). If AIMLAPI was silently quantizing, you'll see the difference immediately.

bash
# Replace KEY values; run both, diff outputs PROMPT='{"model":"claude-opus-4-7","max_tokens":300,"messages":[{"role":"user","content":"Write a haiku about debugging in Python."}]}' curl -s -H "Authorization: Bearer sk-cc-yours" \ -H "Content-Type: application/json" \ -d "$PROMPT" \ https://claudeapi.cheap/api/proxy/v1/chat/completions \ | jq -r '.choices[0].message.content'

Hit your dashboard at /dashboard/usage— the request lands with exact token count and cost. That row is auditable; AIMLAPI's billing logs are not.

Model ID mapping

AIMLAPI uses Anthropic legacy snapshot names. We use the unprefixed canonical names that match what Anthropic ships today.

AIMLAPI ID
ClaudeAPI.cheap ID
claude-3-opus-20240229
claude-opus-4-7
claude-3.5-sonnet
claude-sonnet-4-6
claude-3-haiku-20240307
claude-haiku-4-5
gpt-4-turbo
gpt-5.5
gpt-4o-mini
gpt-5.4-mini
gemini-pro-1.5
gemini-3-pro-preview

Full catalog at /models. We track upstream releases the same week.

FAQ

How do I prove the model isn't being quantized?

Run a fixed determinism-test prompt (low temperature, single-shot, no streaming) against us, then against Anthropic Direct with the same key flow. Token-by-token output should match. We also publish weekly parity benchmark numbers in /trust — they're a public commitment that we never modify model weights.

What if my AIMLAPI balance is non-refundable and I'm stuck?

Two options. Either drain the AIMLAPI balance first on a workload you can afford to spend the lower-quality output on, then cut over to us. Or treat the AIMLAPI deposit as sunk cost and migrate today — Pro is $19 lifetime and pays back fast on real volume. We've seen both paths work.

Is the refund policy actually enforced?

Yes. Email support@claudeapi.cheap within 7 days of a top-up and we refund remaining balance to your originating crypto wallet within 24 hours. The $19 Pro upgrade is also refundable within the same window. Terms are public at /refund-policy.

What does the status page actually track?

status.claudeapi.cheap runs Instatus probes against `/api/proxy/v1/messages` every 60 seconds from multiple regions. Public uptime numbers update in real time. The 99.5% SLA is computed against this exact feed. We can't fake it — Instatus is the third-party source of truth.

Streaming, prompt caching, tool calls — all work?

All three. SSE streaming preserves the full event sequence, prompt caching works for write/read (cache metadata forwarding is intermittent in 2026), tool/function calls round-trip exactly as on Anthropic Direct. We strip upstream vendor fingerprints (billing_transaction_id, system_fingerprint) so your responses stay clean.

Switch to a real proxy — get an API key

No card, crypto only. Basic free forever; $19 lifetime Pro. Read our SLA →