CHANGELOG.md

Changelog

1.11.2 (2026-03-28)

Bug Fixes

• use threading for canary module exec (#294) (4d4a0df)

1.11.1 (2026-03-27)

Bug Fixes

• AE-2322: network volume size bounds, forbid extra args (#290) (6cc7ec2)
• enforce payload size limit and timeout on deserialization (#289) (1240d82)
• idletimeout must be in valid range (#287) (57aebbf)
• lock all workers to Python 3.12, remove multi-version support (#292) (e1f1dd8)
• reject unknown kwargs in @remote (#288) (a651275)
• runtime: handle input=None in generic handler (#286) (cab2d0c)
• stable sorted gpu ids (#285) (759e0f6)

Code Refactoring

• build: replace .flashignore with built-in ignore patterns (#220) (f42c802)
• migrate .runpod directory to .flash (#221) (c6792d6)

1.11.0 (2026-03-25)

Features

• default GPU endpoints to minCudaVersion 12.8 (#277) (e693d7d)
• migrate from AST to import + inspect (#265) (dce6a1f)

1.10.2 (2026-03-23)

Bug Fixes

• deploy: always include env in saveTemplate payload (#281) (69ecac2)
• exclude .env files from deploy artifact by default (#280) (92d0e3e)

1.10.1 (2026-03-20)

Features

• bump max flash tarball size (#258) (4c58a3e)
• cli: add flash update command with passive update check (#237) (5464ce0)

Bug Fixes

• init: flash init show usage menu when called without arguments (#235) (3144d0c)
• resolve .env from CWD and stop _do_deploy env mutation (#236) (583efae)
• retry tarball upload on transient SSL and connection errors (#279) (564f51e)

Code Refactoring

• separate .env from deployed endpoint env vars (#257) (ee26c43)

1.10.0 (2026-03-17)

Features

• enable multi-datacenter support (#266) (5631c32)

1.9.1 (2026-03-12)

Bug Fixes

• auth: clean up API key error message (#269) (7780f28)

1.9.0 (2026-03-12)

Features

• refactor GpuType enum names, add B200 and RTX PRO 6000s (#233) (b709176)

Bug Fixes

• login: prevent env var auth leak in guest sessions (#267) (6316289)

1.8.1 (2026-03-10)

Bug Fixes

• build: remove numpy from dependency blacklist, rename to SIZE_PROHIBITIVE_PACKAGES (#263) (65fde9d)

1.8.0 (2026-03-09)

Features

• Python version-aware Docker image selection (#261) (e10c18e)

1.7.0 (2026-03-05)

Features

• AE-2276: retryable graphql queries (#239) (05a5149)
• auto-select next available port when default is in use (#249) (90276aa)

Bug Fixes

• auto-create app on deploy when app does not exist (#253) (0413e8a)
• CPU instance type not passed through manifest during deploy (#250) (8998123)
• extract dependencies from Endpoint() patterns during build (#252) (3f136bd)
• LB endpoint redeployment desync on second deploy (#251) (6aa3696)

1.6.1 (2026-03-05)

Bug Fixes

• env: handle app-not-found gracefully in flash env list (#243) (97608e7)

Code Refactoring

• Endpoint class as a single entrypoint uniting [@remote](https://github.com/remote) and ServerlessResource-based classes (#223) (5c3f3a6)

1.6.0 (2026-03-04)

Features

• core: replace hardcoded worker values with constants (#201) (08375a1)

Bug Fixes

• deploy: preserve networkVolume config through manifest pipeline (#242) (d0e0a1f)
• stubs: resolve sibling module imports in dependency resolver (#240) (eec4fba)

Code Refactoring

• consolidate HTTP client factories with Content-Type headers (#203) (f52ed84)

1.5.0 (2026-02-27)

Features

• flash login command (#160) (4480308)

Bug Fixes

• add validator for workersmax and workersmin on sls endpoints (#231) (c24a21b)
• dep parsing for remote decorated functions (#230) (f7486cb)
• persist env vars to manifest and deployed endpoints (#226) (4e583c7)

1.4.2 (2026-02-26)

Bug Fixes

• always bundle runpod-flash in flash build + versioning report (#227) (8f883ee)

1.4.1 (2026-02-26)

Bug Fixes

• stubs: resolve in-function @remote imports for cross-endpoint dispatch (#224) (7f168f7)

1.4.0 (2026-02-24)

Features

• fully deployed environment (#215) (8aa4349)

Bug Fixes

• test: align LB activity log mock with Python 3.14 compatible pattern (#218) (cba753e)
• test: stabilize flaky async context manager mock pattern (#216) (1bc369c)

1.3.0 (2026-02-20)

Features

• flash cli polishing treatments (#209) (93bc347)

Code Refactoring

• simplify flash init skeleton for zero-boilerplate flash run (#208) (22894d4)

Documentation

• remove coordinator and mothership terminology from all documentation (#210) (4ba5a36)

1.2.0 (2026-02-17)

Features

• add API key propagation for cross-endpoint calls (#193) (f87c9c1)
• add file-based logging for local CLI usage (#197) (665bcfa)
• add User-Agent header with version, OS, and arch (#202) (5632907)
• AE-2089: update sls endpoint template params (#198) (656fa46)
• cleanup flash deploy/undeploy/build command output format (#191) (c99b486)
• logger: add sensitive data filter to prevent logging API keys and tokens (#200) (10967a4)

Bug Fixes

• docs: change idleTimeout from minutes to seconds (#205) (51693c7)
• prevent false deployment attempts in Flash environments (#192) (f07c9fb)
• runtime: restore on-demand provisioning for flash run (#206) (5859f4b)

Code Refactoring

• remove noisy debug logs from flash (AE-1966) (#204) (826f169)

Documentation

• update CLI documentation for deploy, env, and app commands (#195) (4126b37)

1.1.1 (2026-02-09)

Bug Fixes

• AE-2041: standardize cli cmd names (#179) (9d035e3)

1.1.0 (2026-02-06)

Features

• flash deploy auto builds, automatic app & env management (#178) (cee5e74)
• improve deployment guidance and route discovery (#181) (11681b5)

Performance Improvements

• fail build fast on size limit, remove blocking LB health check (#182) (5278449)

1.0.0 (2026-02-04)

⚠ BREAKING CHANGES

• rename tetra to flash (#165)

Bug Fixes

• 0.25.3 (#173) (98b1ed7)

Code Refactoring

• rename tetra to flash (#165) (7546eaf)

0.25.1 (2026-02-03)

Bug Fixes

• set build id for env before deploying resources (#169) (e5d38d6)

0.25.0 (2026-02-02)

Features

• flash build --preview in local docker for smoke-testing deployments (#167) (a2cc56c)
• code intelligence indexer service for Claude Code (#158) (69b3887)

Bug Fixes

• exclude NetworkVolume from undeploy list command (#164) (b0960d9)

Code Refactoring

• build: remove deprecated handler generator and add verification scripts (#166) (5a253af)
• rename archive.tar.gz to artifact.tar.gz (#170) (1f033c3)

0.24.0 (2026-01-30)

Features

• undeploy resources during flash deploy delete (#155) (5256821)

0.23.1 (2026-01-29)

Bug Fixes

• add flash env id to saveendpoint during flash deploy (#161) (d5c000e)

0.23.0 (2026-01-29)

Features

• disable EU-RO-1 lock in non production environments (5f4772d)
• pull in endpoint base from runpod-python (a970406)
• support specific GPU ids in serverless resource config (#143) (86fde45)

Bug Fixes

• deploy: apply CPU disk auto-sizing to load balancers (#156) (334d582)
• fix usage on VPNs because aiodns is flaky (c55c97b)

0.22.0 (2026-01-27)

Features

• implement cross-endpoint routing reliability infrastructure (#149) (cb6a226)
• implement upfront resource provisioning in deployment flow (#152) (b9c5571)

0.21.0 (2026-01-26)

Features

• add "live-" prefix to live serverless endpoints (#150) (5c8fa33)

0.20.1 (2026-01-22)

Performance Improvements

• tests: make parallel test execution the default (#145) (3faf170)

0.20.0 (2026-01-15)

Features

• add generic handler factory and build integration for Flash (#130) (8c1e6b8)
• add GET /manifest endpoint for mothership service discovery (#139) (2956f09)
• AE-1741 manifest management via gql for flash client (#144) (83979e7)
• build: Add cross-platform build support and package exclusion (#135) (68e0539)
• complete [@remote](https://github.com/remote) support for LoadBalancer endpoints (#131) (f2f34c0)
• cross-endpoint routing for serverless functions (#129) (57ff437)
• mothership manifest sync and caching (#140) (20490ea)
• mothership: implement auto-provisioning with manifest (#136) (14effd4)

Bug Fixes

• drift: Exclude runtime fields from config hash to prevent false positives (#132) (13ced50)

Documentation

• architecture: Add deployment architecture specification (#141) (b1de925)

0.19.0 (2025-12-12)

Features

• AE-1512: deploy() and undeploy() deployable resources directly (#126) (3deac3a)
• cli: Add --auto-provision flag to flash run command (#125) (ee5793c)

Code Refactoring

• use env vars FLASH_HOST and FLASH_PORT (#128) (117a6ae)

0.18.0 (2025-11-28)

Features

• cli: Add flash undeploy command for endpoint management (#121) (cd32ffc)

Code Refactoring

• move endpoint deletion logic to proper abstraction layers (#124) (c253d3b)

0.17.1 (2025-11-19)

Documentation

• update readme for alpha release (#120) (e8093d8)

0.17.0 (2025-11-19)

Features

• add user-friendly error messages for missing RUNPOD_API_KEY (#117) (32dc093)

Documentation

• improved flash init skeleton, contributing, and misc cleanup (#118) (91acf1a)

0.16.1 (2025-11-14)

Bug Fixes

• skeleton: Fix flash init missing hidden files in wheel distributions (#115) (c3bf137)

0.16.0 (2025-11-14)

Features

• cli: Add flash init with project skeleton template and in-place initialization (#110) (155d6ee)

0.15.0 (2025-11-14)

Features

• client: add async function support to remote decorator (#112) (b0222e0)
• resources: Support for Serverless.type QB|LB (#109) (6e63459)

0.14.0 (2025-10-31)

Features

• AE-1202 add flash cli cmd init and run (#96) (75b2baf)

0.13.0 (2025-10-09)

Features

• Command Line Interface (#50) (5f4cde6)

Bug Fixes

• repair build and release actions (#97) (4a73fa8)

Code Refactoring

• deprecate hf_models_to_cache (#95) (963bfd7)

0.12.0 (2025-09-15)

Features

• better clarity on provisioning CPU Serverless Endpoints (#88) (efec224)
• thread-safe resource provisions and remote executions (#91) (440b36f)

Bug Fixes

• download accelerator changes broken regular endpoint invocations (#86) (759f996)

0.11.0 (2025-08-19)

Features

• Add download acceleration for dependencies and HuggingFace models (#83) (e47c9e3)

0.10.0 (2025-08-07)

Features

• Add idempotent network volume deployment (#79) (289d333)

0.9.0 (2025-08-04)

Features

• AE-961 Add class serialization caching for remote execution (#76) (95f9eed)

0.8.0 (2025-07-22)

Features

• AE-815 Add class based execution (#72) (d2a70ad)

0.7.0 (2025-07-21)

Features

• AE-892 - Lock the network volume region (#67) (2b7a3ea)

Bug Fixes

• downgrade release-please to v3 for simplicity (#70) (06f05d1)
• use default release-please PR title pattern (#68) (786df98)

0.6.0 (2025-07-10)

Features

• AE-811 Add a class to create network volume using REST (abb82b7)

0.5.5 (2025-07-10)

Bug Fixes

• remove sigstore signing to simplify PyPI publishing (65037c5)

0.5.4 (2025-07-10)

Bug Fixes

• revert to simple working CI/release workflows, remove broken shared workflow (3fe4cf0)
• separate sigstore artifacts from dist/ to avoid PyPI validation errors (6b0038e)

Documentation

• add contributing section to README with release system documentation reference (8b87b9c)

0.5.3 (2025-07-10)

Bug Fixes

• update sigstore action to v3.0.0 to avoid deprecated upload-artifact@v3 (92a6891)

0.5.2 (2025-07-10)

Bug Fixes

• revert to simple working CI/release workflows, remove broken shared workflow (62dab19)

0.5.1 (2025-07-10)

Bug Fixes

• adjust release-please title pattern to match existing PR #55 (091dbff)
• only upload artifacts to release when tag_name is available (e8bdcf1)
• revert to clean release-please title pattern for future releases (c0171ca)
• still seeing issues with the CI and Release (#60) (81df31a)

0.5.0 (2025-07-09)

Features

• pytest with unit and integration (19dcac9)

Bug Fixes

• cannot do both (c64d895)
• configure release-please to handle all commit types (#57) (7de5c8d)
• correction (9982236)
• Pydantic's deprecation warning of class Config (d1f18f2)
• ruff cleanup (8251597)
• ruff linting (6d8e87c)

Documentation

• pull request template (#58) (0377fa4)

0.4.2 (2025-06-26)

Bug Fixes

• consolidate PodTemplate overrides with the defaults per ServerlessResource (2b8dc16)
• template overrides (0fd6429)

0.4.1 (2025-06-26)

Bug Fixes

• uv publish (6c4f9a9)
• uv publish (55ae7c0)

0.4.0 (2025-06-26)

Features

• CPU Endpoints and Live Serverless (c335fd1)

Bug Fixes

• CpuInstanceType values (9cf958b)

0.3.0 (2025-06-23)

Features

• added diffusers as examples group dependency (e330bf9)
• AE-392: serverless execute function calls runpod SDK (43900e5)
• AE-394: examples updated (27c3afb)
• AE-394: examples updated (d2bfe1b)
• AE-394: LiveServerless that locks the template id (d1032a6)
• AE-394: LiveServerless that locks the template id (1ba5bfc)
• AE-394: resilient serverless calls with retries and auto-cancellations (78bb250)
• AE-394: retry, backoff, on cancel appropriately (4b75941)
• AE-394: retry, backoff, on cancel appropriately (96d1001)
• AE-432: Logging (a93f88e)
• AE-432: Logging (718de18)
• AE-442: Docker image template polish + cleanup (75471b2)
• AE-470: shadow deploy as runpod_flash (749d427)
• AE-494: remove all docker- worker-related content (72a4b9a)
• AE-494: removed any worker-related content (72464d1)
• AE-517: logs from remote printed as Remote | <log> (d8f4ee1)
• AE-517: logs from remote printed as Remote | <log> (ab1d729)
• AE-517: remote error logs are piped back (b668e3e)
• AE-517: remote error logs are piped back (4c5bf4c)
• AE-565 singledispatch stub_resource for LiveServerless (5fe2de7)
• incremental changes, starting with deployments (1ce1b3b)
• is_deployed() to check and redeploy in case endpoint is inaccessible (4aa0932)
• Prefer A40 for GPU list when "any" is set (6a60418)
• preparing json.dump of complex objects with normalize_for_json (48cedd6)
• Pydantic modeling of Runpod resources (87067bb)
• remote stub for ServerlessEndpoint calls (739aa93)
• resilient redeployments (fd2e78d)
• resilient serverless calls with retries and auto-cancellations when unhealthy/throttled (9f53e8c)
• resource utility, inquire to map runpod calls to pydantic models (096239f)
• ServerlessEndpoint inherits ServerlessResource (c0e64f9)
• ServerlessResource.run_sync (33e3fff)
• SingletonMixin for ResourceManager and RemoteExecutionClient (e0ae91b)
• Support for ServerlessEndpoint remote calls (e53f3aa)

Bug Fixes

• add missing arg to request (21d4e0c)
• add missing arg to request (656150f)
• AE-392: deployable serverless corrections (57b1cd2)
• AE-392: deployable serverless corrections (74bd17d)
• cleanup from ruff suggestions (61048d2)
• cleanup project setup (9ed50ab)
• cleanup unused code (ad78de7)
• implicit load_dotenv() before everything (ff0030b)
• implicit load_dotenv() before everything (2d3c710)
• make examples (7de342c)
• missed these commits (5ed3a8a)
• passing GPU names aren't supported yet (3798ac0)
• Put A40s in front of the list when gpus=any (717efb1)
• python 3.9 compatible (5ddfd09)
• release-please bugs (fd6d02e)
• release-please bugs (76b3eee)
• ruff cleanup (739adaf)
• simplify make dev that syncs and installs examples (d1a3e62)
• try catch for matrix example (f7a5012)
• use of pathlib.Path where appropriate (4ee717c)

0.2.1 (2025-06-23)

Bug Fixes

• implicit load_dotenv() before everything (2d3c710)