Crypto news in. Model-ready sentiment out.
One API call turns any headline or article into a structured sentiment score your prediction models can use — in under 60ms.
100 free credits on signup · No credit card required
Request
"text": "Bitcoin ETF approval sparks record institutional inflows"
Response
<60ms
Inference latency
89%
Model accuracy
99.9%
API uptime
How it works
Three steps to structured sentiment
No ML expertise required. Send text, get a score. It's that simple.
Send your text
POST any crypto headline, tweet, or article to a single endpoint. Raw text in — up to 140 characters.
POST /v1/sentiment
{ "text": "SEC hints at BTC ETF..." }We extract the signal
Our crypto-tuned NLP model classifies sentiment and scores confidence in under 60ms.
Classifying: sentiment, score, confidence
You get structured output
A sentiment label, a score from 0 to 1, and a confidence level — ready for your feature vector.
{ "sentiment": "positive",
"score": 0.82,
"confidence": "high" }Inside the model
Meet bb-sentiment-1, our latest production release
bb-sentiment-1 is our current flagship release, tuned to turn headlines, articles, and fast-moving market language into dependable sentiment signals for research and trading systems.
Tech specs
Accuracy
89%
Current validation benchmark
Latency
<60ms
Average inference response
Parameters
>400M
Number of parameters in the model
Dataset
Large curated corpus
Headlines, articles, and crypto-native market language.
Validation
Advanced release gates
Holdouts, filtering, and repeated regression checks.
Architecture
Modern BERT methods
Context-aware encoder training for nuanced sentiment.
Release notes
bb-sentiment-1
Large dataset
Trained on a large, curated dataset of crypto headlines, news stories, and market commentary so the model learns both short-form catalysts and longer contextual narratives.
Advanced validation
Advanced validation combines strict holdouts, noisy-sample screening, and repeated regression checks across multiple text formats before a release is promoted.
Modern BERT methods
Built with modern BERT-style encoder methods so the model captures context, negation, tone shifts, and domain-specific phrasing more reliably than keyword-driven sentiment pipelines.
The problem we solve
Raw text in. Market signal out.
Stop building sentiment models. Start using sentiment data.
“SEC Chair hints at Bitcoin regulation changes amid growing institutional pressure”
Unstructured text. No numeric signal. Can't feed this to a model.
Same headline, one API call later:
Features
Designed for prediction models
Clean inputs, predictable outputs, and infrastructure that stays fast when markets get noisy.
Sub-60ms inference
<60msavg responseLow enough latency for real-time trading decisions and live data pipelines. No batching required.
Crypto-native model
89%accuracyTrained on crypto market language — ETF approvals, halvings, liquidations, regulatory nuance.
One endpoint, one key
1endpointJSON in, JSON out. Integrate in minutes. No SDKs, no dependencies, no configuration.
Burst-ready infrastructure
99.9%uptimeScales automatically during market events. Your sentiment pipeline handles volume spikes.
Use cases
Built for teams that move markets
From solo quants to institutional platforms — structured sentiment fits wherever you need it.
Algorithmic trading
Add sentiment as a real-time feature in your signals. Gate entries and exits on headline tone.
Hedge fund research
Enrich pipelines with structured sentiment from news sources across the entire crypto market.
Trading bots
Add a news-aware layer to automated strategies. Check sentiment before every execution.
Dashboards & monitoring
Build tools that track real-time sentiment across assets alongside price and volume.
Pricing
Simple, credit-based pricing
Pay only for what you use. No subscriptions, no minimums, no surprises.
Credit-based usage
10 credits per inference · 1 credit for cached results
10
credits / inference
100
free credits on signup
$0
monthly minimum
Starter
$10
500 credits · $0.020/credit
Growth
$40
2,500 credits · $0.016/credit
Custom — 5,000+ credits
Pick your amount, pay less per credit as you scale.
$0.012/credit
5,000–9,999
$0.010/credit
10,000–24,999
$0.008/credit
25,000–49,999
$0.006/credit
50,000+
- 100 free credits on signup — no credit card required
- Credits never expire
- Full API access from day one
- Buy more credits anytime as you scale
Need enterprise volume? Contact us
Get started today
Your model is only as good as its inputs.
Add crypto sentiment as a feature in minutes. Every signup gets 100 free credits — no credit card required.