NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI, #3569 most downloaded on PyPI
The official Python library for the fireworks API
Latest stable
1.2.33
Last release 8 days ago
08 Nov 2026
Ships on a steady schedule
a new release about every 2 weeks
Some releases are documented
notes for 34 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
4 years old
206 releases · first in 2022
Drag across the ruler to compare, or tap a release to open it.
## 1.0.1-alpha.63 (2026-04-22) Full Changelog: v1.0.0-alpha.63...v1.0.1-alpha.63
Full Changelog: v1.0.0-alpha.63...v1.0.1-alpha.63
downstream compat shims, CI gates, and test coverage
Full Changelog: v1.0.0-alpha.62...v1.0.0-alpha.63
Your coding agent can read these notes before it upgrades. Set up the MCP server →
ensure file data are only sent as 1 parameter
Full Changelog: v1.0.0-alpha.61...v1.0.0-alpha.62
sdk: deployment update, delta-chain reset, and stale-error handling for re-attach
Full Changelog: v1.0.0-alpha.60...v1.0.0-alpha.61
training-sdk: add create_base_training_client for LoRA self-reference
Full Changelog: v1.0.0-alpha.59...v1.0.0-alpha.60
deprecate hot_load_deployment_id with runtime warning
training-sdk: disable delta hotload for LoRA adapters
## 1.0.0-alpha.57 (2026-04-03) Full Changelog: v1.0.0-alpha.56...v1.0.0-alpha.57
Full Changelog: v1.0.0-alpha.56...v1.0.0-alpha.57
add skip_validations to TrainerJobConfig
support versioned paths in resolve_training_profile
training-sdk: close dogfooding gaps for local loops
deployment_shape property returns versioned path
add hot_load_trainer_job to DeploymentConfig
restore missing urlencode import in trainer.py
internal: implement indices array format for query and form serialization
Full Changelog: v1.0.0-alpha.49...v1.0.0-alpha.50
streaming-only inference with adaptive concurrency
sdk: add max_concurrency to DeploymentSampler
Full Changelog: v1.0.0-alpha.47...v1.0.0-alpha.48
sdk: auto-resolve account_id from API key
training-sdk: add promote_checkpoint, strict shape validation, and API key remapping
re-apply GradAccNormalization enum and optim_step override
Full Changelog: v1.0.0-alpha.44...v1.0.0-alpha.45
by_alias unless set (39e4145)## 1.0.0-alpha.44 (2026-03-16) Full Changelog: v1.0.0-alpha.43...v1.0.0-alpha.44
Full Changelog: v1.0.0-alpha.43...v1.0.0-alpha.44
add grad_accumulation_normalization to optim_step
sdk: async httpx with n=1 completions
training-sdk: centralize HTTP transport in _RestClient base class
training-sdk: bump tinker to 0.15.0
training-sdk: two-path shape override contract
trainer: route via API gateway instead of directRouteHandle
Full Changelog: v1.0.0-alpha.37...v1.0.0-alpha.38
sdk: apply_shape always autofills from training shape values
Full Changelog: v1.0.0-alpha.36...v1.0.0-alpha.37
extra_items from PEP 728 (2a492a9)training shape apply_shape, tinker pin, and README cleanup
make training extras publishable on PyPI
add client-side max seq length filter and align defaults
Full Changelog: v1.0.0-alpha.28...v1.0.0-alpha.29
test_proxy_environment_variables more resilient to env (6de8f74)add R3 routing matrix side-channel injection for official tinker compat
Full Changelog: v1.0.0-alpha.27...v1.0.0-alpha.28
test_proxy_environment_variables more resilient (6b909f0)Add Anthropic-compatible Messages API documentation
Full Changelog: v1.0.0-alpha.26...v1.0.0-alpha.27
client: add custom JSON encoder for extended type support
## 1.0.0-alpha.25 (2026-01-27) Full Changelog: v1.0.0-alpha.24...v1.0.0-alpha.25 ### Features * update API reference
Full Changelog: v1.0.0-alpha.24...v1.0.0-alpha.25
ci: upgrade actions/github-script
Full Changelog: v1.0.0-alpha.23...v1.0.0-alpha.24
actions/github-script (7869665)validation: add Pydantic models and validation for dataset rows in training workflow
Full Changelog: v1.0.0-alpha.22...v1.0.0-alpha.23
actions/checkout version (14881d2)Add model_kind and model_state schemas to stainless.yml
Full Changelog: v1.0.0-alpha.21...v1.0.0-alpha.22
internal: codegen related update
Full Changelog: v1.0.0-alpha.20...v1.0.0-alpha.21
Remove deprecated Evaluator fields
Full Changelog: v1.0.0-alpha.19...v1.0.0-alpha.20
Document GLM 4.7 / reasoning_history
Full Changelog: v1.0.0-alpha.18...v1.0.0-alpha.19
## 1.0.0-alpha.18 (2025-12-20) Full Changelog: v1.0.0-alpha.17...v1.0.0-alpha.18
Full Changelog: v1.0.0-alpha.17...v1.0.0-alpha.18
## 1.0.0-alpha.17 (2025-12-20) Full Changelog: v1.0.0-alpha.16...v1.0.0-alpha.17
Full Changelog: v1.0.0-alpha.16...v1.0.0-alpha.17
## 1.0.0-alpha.16 (2025-12-19) Full Changelog: v1.0.0-alpha.15...v1.0.0-alpha.16 ### Features * Add deployment shapes to sdk
Full Changelog: v1.0.0-alpha.15...v1.0.0-alpha.16
run "make" + add SDK generation guide
Full Changelog: v1.0.0-alpha.14...v1.0.0-alpha.15
## 1.0.0-alpha.14 (2025-12-19) Full Changelog: v1.0.0-alpha.13...v1.0.0-alpha.14 ### Documentation * add more examples
Full Changelog: v1.0.0-alpha.13...v1.0.0-alpha.14
Updated stainless.yml to include execute train step
Full Changelog: v1.0.0-alpha.12...v1.0.0-alpha.13
stainless.yml to include execute train step (617aaf9)--fix argument to lint script (859cb4f)use async_to_httpx_files in patch method
Full Changelog: v1.0.0-alpha.11...v1.0.0-alpha.12
## 1.0.0-alpha.11 (2025-12-16) Full Changelog: v1.0.0-alpha.10...v1.0.0-alpha.11 ### Chores * speedup initial import
Full Changelog: v1.0.0-alpha.10...v1.0.0-alpha.11
Expose Create Evaluator method as part of Docs/SDK
Full Changelog: v1.0.0-alpha.9...v1.0.0-alpha.10
## 1.0.0-alpha.9 (2025-12-12) Full Changelog: v1.0.0-alpha.8...v1.0.0-alpha.9
Full Changelog: v1.0.0-alpha.8...v1.0.0-alpha.9
## 1.0.0-alpha.8 (2025-12-12) Full Changelog: v1.0.0-alpha.7...v1.0.0-alpha.8 ### Features * add pagination (d2ab4bc) * Automate SDK Pagination Config
## 1.0.0-alpha.7 (2025-12-10) Full Changelog: v1.0.0-alpha.6...v1.0.0-alpha.7 ### Features * Remove methods from SDK
Full Changelog: v1.0.0-alpha.6...v1.0.0-alpha.7
add evaluator job / evaluator CRUD APIs to API/SDK
Full Changelog: v1.0.0-alpha.5...v1.0.0-alpha.6
types: allow pyright to infer TypedDict types within SequenceNotStr
Full Changelog: v1.0.0-alpha.4...v1.0.0-alpha.5
remove external reference comments from example file
Full Changelog: v1.0.0-alpha.3...v1.0.0-alpha.4
add account_id configuration section and update examples
Full Changelog: v1.0.0-alpha.2...v1.0.0-alpha.3
add dataset upload documentation with complete workflow examples
Add documentation about streaming usage in generated README for Python SDK
Full Changelog: v0.0.1...v1.0.0-alpha.1
model_dump and model_dump_json for Pydantic v1 (e7405ba)Nothing published for this version
One column per quarter.