For developers
AI inference, priced to undercut hyperscalers for open-source models.
OpenAI-compatible API. Drop-in replacement. No lock-in. We take any idle GPU and route AI workloads to it intelligently.
Sign in
API key
sk_live_9pX2…wN4k✓ copied
Llama-3-70B-Instruct
popular
Mixtral 8×7B
Llama-3-8B-Instruct
fast
FLUX.1 [dev]
new
SDXL 1.0
Whisper-large-v3
# Before — OpenAI
base_url="https://api.openai.com"
# After — Wattlend (just one line)
base_url="https://api.wattlend.com"
✓ Same SDK · Same request shape · Same streaming format
Routed to host-garage-rigus-east-1 · 47ms
The decentralized AI cloud aggregates idle GPUs and routes inference to the best-fit node in real time▍
TTFT
126 ms
Tok/s
142
Reliability
98.4
This month$23.41
Llama-3-70B input tokens38.4M$21.12
Llama-3-70B output tokens1.9M$2.28
Egress (free under 100GB)14GB$0.00
vs AWS Bedrock equivalent−47%
STEP 01 / 05auto-advance
1. Sign up — get an API key in 30 seconds
Email or Google. No credit card up front. Generate an API key with one click; it works against an OpenAI-compatible endpoint.
Honest pricing vs hyperscalers
Per million tokens, USD. We re-verify these every quarter.
| Model | Wattlend (in / out) | AWS Bedrock (in / out) | You save |
|---|---|---|---|
| Llama-3-8B-Instruct | $0.20 / $0.40 | $0.30 / $0.60 | ~35% |
| Llama-3-70B-Instruct | $0.55 / $1.20 | $0.99 / $2.50 | ~50% |
| Mixtral 8×7B | $0.25 / $0.50 | $0.40 / $0.80 | ~38% |
| SDXL 1024×1024 | $0.0035 / image | $0.0080 / image | ~56% |
| Whisper-large-v3 | $0.0030 / min | $0.0060 / min | ~50% |
Pay-as-you-go from the first request — no minimums, no monthly commitments. See /pricing for the full table, including where we're more expensive (we're honest about that too).