Jesse Inference API
Sub-millisecond memory-lookup inference engine for CAD kernels, gaming simulations, formal logic, and real-time reasoning. Open Paid Beta with unlimited daily volume, strict 1 request per 2 seconds rate enforcement, $20/month commercial developer subscription via Stripe, and dual-model continuous learning.
It remembers what you tell it — specific to your API key. Any memories, factual assertions, and conversational states persist strictly partitioned to your individual API key.
We encourage data/IP caution during this open beta. Please exercise standard intellectual property caution when submitting sensitive or proprietary material during the beta period.
We do not, and will not, pool RSI learning across all user interactions. Learning is pooled for that user specifically in combination with our frontier models—your data and learned facts are never shared or cross-pollinated into other users' models.
-
[+]
Open Paid Beta Access — instant API key provisioning, priority DGX Blackwell compute, and live Lemma Bank streaming.
-
[+]
Unlimited Daily Requests — no daily or monthly token quotas (up to 43,200 req/day).
-
[+]
Strict 1 req / 2s Rate Limit — predictable sustained throughput; returns 429 on sub-2s bursts.
-
[+]
Dual Model Architecture — toggle
jesse-prod(continuous learning) orjesse-pristine(immutable baseline). -
[+]
OpenAI API Drop-In Format — native support for
POST /api/v1/chat/completionswith streaming and tool calling. -
[+]
Key-Isolated RSI Learning — remembers what you tell it specific to your API key; learning is never pooled across users.
-
[+]
Zero-Weight Sub-Millisecond Speed — 5.1 µs RAM lookup, zero cold-start GPU spinup.
Authorization: Bearer <key>.Key-Specific Memory: It remembers what you tell it — specific to your API key. We do not, and will not, pool RSI learning across all user interactions; it is pooled for that user specifically in combination with our frontier. We encourage data/IP caution during this open beta.
from openai import OpenAI
# Initialize client pointing to Jesse at jesse.solidsf.com (or jesse.my)
client = OpenAI(
base_url="https://jesse.solidsf.com/api/v1", # or https://api.jesse.my/v1
api_key="jesse_live_YOUR_KEY_HERE"
)
# Query jesse-prod (continuous learner) or jesse-pristine (immutable baseline)
response = client.chat.completions.create(
model="jesse-prod",
messages=[
{"role": "system", "content": "You are Jesse, autonomous engineering agent."},
{"role": "user", "content": "How do Euler-Poincaré invariants prevent BRep failures?"}
],
temperature=0.0
)
print(response.choices[0].message.content)
Handles live customer production requests. Dynamically indexes validated user assertions, feedback corrections, and domain-specific terminology strictly partitioned to your API key. Learning is pooled for that user specifically in combination with our frontier—never pooled across different users.
Completely isolated from user traffic. Strictly evaluated against deterministic test suites (LiveBench, GPQA, KataGo, ARC, HumanEval) to guarantee 0 regressions.
• No Cross-Tenant Pooling: We do not, and will not, pool RSI learning across all user interactions. It is pooled for that user specifically in combination with our frontier.
• Open Beta IP Caution: We encourage data/IP caution during this open beta.
GET /api/lemmas every 15s • Type-checked by Lean 4 kernel with 0 axioms added
| Method | Endpoint | Description | Rate Limit |
|---|---|---|---|
| POST | /api/v1/chat/completions |
OpenAI-compatible chat completions (supports streaming, tools, and model choice) | 1 req / 2s |
| GET | /api/v1/models |
List available models (jesse-prod, jesse-pristine) and active learning stats |
1 req / 2s |
| POST | /api/v1/feedback |
Submit thumbs up/down and factual corrections directly to continuous learning pool | 1 req / 2s |
| GET | /api/v1/learnings/stats |
Inspect telemetry on collected user interactions, buffered corrections, and merge queue | 1 req / 2s |
| POST | /api/admin/consolidate |
Trigger distillation and regression check to merge prod learnings into candidate pristine release | Admin |
| POST | /api/billing/checkout |
Create Stripe Checkout session for $20/month Developer subscription | Standard |
| POST | /api/billing/portal |
Generate Stripe Customer Billing Portal URL for subscription management | Standard |
| POST | /api/billing/webhook |
Stripe webhook receiver handling automated key provisioning and subscription changes | Unrestricted |
| GET | /api/lemmas |
Live stream of Lean 4 kernel-certified theorems, proofs, and structural bounds across Millennium domains | Standard |