Skip to main content

​
1.1.0 — 2026-05

Token-based metering
  • New tex.usage.today() and tex.usage.summary(month=...) methods.
  • Every remember and recall response now carries a usage field with tokens_in and tokens_out.
  • Daily quotas enforced at the engine. Exceeding them raises RateLimitError.
Conversation-native verbs
  • tex.conversations.remember(turns=..., session_id=...) for first-class turn ingestion.
  • tex.recall(q=..., session_id=..., mode="active"|"deep") with confidence scoring and optional timeline.
Improvements
  • HTTP/2 enabled by default for ~25% lower tail latency on recall.
  • Automatic JWT refresh on 401, with single-retry semantics.
  • Stable type stubs (py.typed) shipped with the wheel.

​
1.0.0 — 2026-03

Initial public release. Supermemory-compatible verbs (add, search, profile).

​
Roadmap

Async client

Native async methods on an AsyncTex class. Same methods, but awaitable. Targeting Q3.

Per-call user scoping

tex.recall(..., user_id=...) without constructing a new client. Targeting 1.2.

TypeScript SDK

First-class TS client mirroring the Python API. Targeting Q3.

Webhook ingestion

Stream turns from your event bus directly into memory. Targeting Q4.

Server-side quota alerts

Email notifications at 80% / 100% of either daily cap. Targeting 1.2.

Scoped API keys

Separate read, write, and usage:read scopes so each key does only what it should.