//

log

changelog

//

log

changelog

Changelog

Changelog

Every release, in the order it shipped. Breaking changes are marked − and always come with a migration note.

Every release, in the order it shipped. Breaking changes are marked − and always come with a migration note.

v2.4.0

Feature

Suites now fan out across up to 64 workers. A 1,200-case suite that took 9 minutes finishes in 41 seconds.

+

--workers flag and per-suite worker limits

~

Default concurrency is now 8 (was 1)

v2.3.0

Feature

Write graders that look things up: check an order in your staging API before deciding a refund answer was right.

+

tool() helper for graders, with a 2s timeout

~

Grader logs now show every tool call

v2.2.2

Improvement

The CLI no longer downloads model metadata on every run. First run in a fresh CI container is 3.1s faster.

+

Local cache in .tessel/

−

--no-cache flag (the cache is safe to keep)

v2.2.0

Feature

Cases that flip between pass and fail on the same input are marked flaky and kept out of the merge gate.

+

flaky: true marker in results

~

Threshold math ignores flaky cases

v2.1.4

Fix

Case files with accented keys failed to parse on Windows runners. They now load the same everywhere.

~

YAML parser updated to the 1.2 core schema

v2.1.0

Feature

The Python SDK is stable. Same API as Node, typed, async-ready.

+

tessel on PyPI, sync and async clients

−

Beta import path tessel_beta

v2.4.0

Feature

Suites now fan out across up to 64 workers. A 1,200-case suite that took 9 minutes finishes in 41 seconds.

+

--workers flag and per-suite worker limits

~

Default concurrency is now 8 (was 1)

v2.3.0

Feature

Write graders that look things up: check an order in your staging API before deciding a refund answer was right.

+

tool() helper for graders, with a 2s timeout

~

Grader logs now show every tool call

v2.2.2

Improvement

The CLI no longer downloads model metadata on every run. First run in a fresh CI container is 3.1s faster.

+

Local cache in .tessel/

−

--no-cache flag (the cache is safe to keep)

v2.2.0

Feature

Cases that flip between pass and fail on the same input are marked flaky and kept out of the merge gate.

+

flaky: true marker in results

~

Threshold math ignores flaky cases

v2.1.4

Fix

Case files with accented keys failed to parse on Windows runners. They now load the same everywhere.

~

YAML parser updated to the 1.2 core schema

v2.1.0

Feature

The Python SDK is stable. Same API as Node, typed, async-ready.

+

tessel on PyPI, sync and async clients

−

Beta import path tessel_beta

tessel

Evals for LLM apps, on every commit.

Release notes, once a month

All systems normal · © 2026 Tessel Labs

tessel

tessel

Evals for LLM apps, on every commit.

Release notes, once a month

All systems normal · © 2026 Tessel Labs

tessel

Create a free website with Framer, the website builder loved by startups, designers and agencies.