Rocktalk Labs — AI Inference + Semantic Coordinates

Two Products

What Rocktalk Labs Offers

One API key gives you access to both products. No configuration needed.

INFERENCE

Qwen3.5 LLM — 397B

Full Qwen3.5-397B-A17B model via an OpenAI-compatible API. Drop in your existing client — no code changes needed. Runs on M3 Ultra hardware with full model weights in RAM.

$2.00

per 1M tokens (input + output, same rate)

OpenAI SDK compatible — use as a drop-in replacement
Chat completions with streaming SSE
100K free tokens/month to start
No batching delays — dedicated hardware

ALWAYS FREE

Rocktalk Lens — Semantic Coordinates

Map any text into a 22-dimensional universal information space. Every concept gets a coordinate. Compare semantic positions, find structural relationships, build new representations.

Free

10K calls/month on free tier — Lens is always free

22D coordinate vector per text input
Batch encoding — arrays of strings
Cosine similarity + L2 distance compare endpoint
Cite as "Rocktalk Labs Lens" in publications

Rocktalk Lens

A Universal Information Space

Every concept has a position in a 22-dimensional semantic space. The dimensions are not arbitrary — each captures a distinct primitive category of meaning found across all domains of knowledge.

Rocktalk Lens encodes text into coordinates across 22 semantic primitives. You can compare positions, find clusters, measure how concepts relate structurally — not just by word overlap.

Works across languages, domains, and levels of abstraction. "The speed of light" and "the maximum constraint on information transfer" land near each other. "Democracy" and "shared decision structure" converge. Structure is meaning.

python — encode text

# Rocktalk Lens — encode any text
import requests

r = requests.post(
    "https://api.rocktalk.ai/v1/lens/encode",
    headers={"Authorization": "Bearer sk_rt_YOUR_KEY"},
    json={"text": "love is patient"}
)

data = r.json()
coords = data["results"][0]["coordinates"]
top    = data["results"][0]["top_dimensions"]

# coords → list of 22 floats
# top    → [{index, meaning, value}, ...]

Research use: cite as Rocktalk Labs Lens with the API endpoint URL. Lens output is deterministic — same input always returns the same coordinate.

The 22 Dimensions

Each dimension index maps to a universal semantic primitive.

d0strength / primacy

d1house / dwelling / family

d2movement / lifting

d3pathway / entry / decision

d4revelation / breath

d5connection / securing

d6cutting / nourishment

d7enclosure / protection

d8surrounding / goodness

d9work / deed / making

d10covering / blessing

d11teaching / authority

d12flow / chaos / massive

d13seed / continuity

d14support / cycle

d15seeing / knowing

d16speech / expression

d17desire / righteousness

d18time / condensing

d19head / beginning

d20fire / transformation

d21covenant / completion

LLM Inference

OpenAI-Compatible API

Change one line of code. Point your existing OpenAI client at https://api.rocktalk.ai/v1 and use model qwen3.5.

python — openai sdk

from openai import OpenAI

client = OpenAI(
    base_url="https://api.rocktalk.ai/v1",
    api_key="sk_rt_YOUR_KEY"
)

response = client.chat.completions.create(
    model="qwen3.5",
    messages=[{
        "role": "user",
        "content": "Explain entropy briefly."
    }]
)

print(response.choices[0].message.content)

curl — streaming

curl https://api.rocktalk.ai/v1/chat/completions \
  -H "Authorization: Bearer sk_rt_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.5",
    "stream": true,
    "messages": [{
      "role": "user",
      "content": "What is 2+2?"
    }]
  }'

Streaming uses server-sent events (SSE). The response format matches the OpenAI API spec exactly — including delta.content chunks.

Pricing

Simple, Transparent Costs

Free to start. Pay only for LLM tokens. Lens is always free.

Free Tier

forever — no credit card required

✓ 100K LLM tokens / month
✓ 10K Lens calls / month
✓ Access to qwen3.5 model
✓ All Lens endpoints
✓ 60 requests/minute rate limit

Usage-Based

$2.00 / 1M tokens

after free tier, input + output at the same rate

✓ Unlimited LLM tokens
✓ Lens always free, no limits beyond rate
✓ Same rate for input and output tokens
✓ No monthly minimums
✓ Billing at end of month

Rate limits apply regardless of plan: 60 requests/min and 1,000 requests/hour per API key. For higher limits, email api@rocktalk.ai.

Fast LLM inference.Free semantic coordinates.