PyPI · #3732 most downloaded on PyPI

fireworks-ai

1.2.191.2M downloads/mofw-ai-external/python-sdk ↗New releases in your daily digest, changelog included.

The official Python library for the fireworks API

Last release 2 days ago
02 Oct 2026
Ships on a steady schedule
a new release about every 2 weeks
Some releases are documented
notes for 20 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
4 years old
192 releases · first in 2022

Release timeline

192 releases since 2022

20222023202420252026

20222026

One column per quarter.

Releases

1.2.192 Oct 20262 sources agree
training-sdk: automatically select RDMA weight sync only for dedicated full-parameter trainers and inference replicas that all advertise support; olde
1.2.19 (2026-10-02)

Full Changelog: v1.2.18...v1.2.19

Features

training-sdk: automatically select RDMA weight sync only for dedicated full-parameter trainers and inference replicas that all advertise support; older images, LoRA, and deployments using gateway shard fan-out retain save-weights followed by file hotload

datasets: add fireworks.lib.dataset_upload.upload_dataset_shards to upload a .jsonl file or a directory of top-level .jsonl shards as one dataset (single upload-endpoint registration, credential-free signed-URL PUTs, then validate_upload), plus collect_dataset_shards and count_dataset_examples helpers; fails closed for CMEK datasets
Open source →
1.2.181 Oct 20262 sources agree
training-sdk: add forward-only reads for per-token projection outputs
1.2.18 (2026-10-01)

Full Changelog: v1.2.17...v1.2.18

Features

training-sdk: add forward-only reads for per-token projection outputs

training-sdk: expose supports_router_replay on training clients, reported by the trainer at create_model, so callers can decide Router Replay without fetching the base model (None when the trainer does not report it)

training-sdk: raise ModelDetailsUnavailableError (a RuntimeError subclass carrying status_code) from FireworksClient.model_is_moe when the model record cannot be read, so callers can tell an inaccessible model (403/404) from a control-plane failure

Bug Fixes

training-sdk: make projection custom-loss helpers safe to call from synchronous and asynchronous code

1.2.1728 Sept 20262 sources agree

training-sdk: add lora_init_method to create_training_client ( init_method on create_lora_training_client ) to pick each model's LoRA initialization a

1.2.17 (2026-09-26)

Full Changelog: v1.2.16...v1.2.17

Features

training-sdk: add lora_init_method to create_training_client (init_method on create_lora_training_client) to pick each model's LoRA initialization at create time: "kaiming" (default) or "nora" (NoRA-init)

Open source →

1.2.1625 Sept 20262 sources agree

training-sdk: add projection_head_dim to managed service and training-client creation for train-only projection heads, and output="projection" for for

1.2.16 (2026-09-25)

Full Changelog: v1.2.15...v1.2.16

Features

training-sdk: add projection_head_dim to managed service and training-client creation for train-only projection heads, and output="projection" for forward_backward_custom gradients with respect to projection outputs

Open source →

1.2.1522 Sept 20262 sources agree

training-sdk: retain per-token top-logprob token IDs and values in sampling results and TITO trajectory artifacts

1.2.15 (2026-09-22)

Full Changelog: v1.2.14...v1.2.15

Features

training-sdk: retain per-token top-logprob token IDs and values in sampling results and TITO trajectory artifacts

Bug Fixes

training-sdk: preserve multidimensional custom-loss gradients as shaped tensor data during backward requests
training-sdk: preserve failed trainer status when reporting provisioning errors to managed training jobs
training-sdk: upgrade pyqwest to 0.10.0 while preserving system TLS roots

Open source →

1.2.1418 Sept 20262 sources agree

training-sdk: select Parquet R3 using a small completion request; older inference servers retain inline routing without a capability endpoint

1.2.14 (2026-09-18)

Full Changelog: v1.2.13...v1.2.14

Features

training-sdk: select Parquet R3 using a small completion request; older inference servers retain inline routing without a capability endpoint
training-sdk: allow sampling requests to set r3_ttl_seconds for shared Parquet routing files; the default retention is six hours

Open source →

1.2.1318 Sept 20262 sources agree

deployments: add accept_shapeless_risk opt-out on deployments.create for intentionally creating a deployment without a pre-validated shape; document t

1.2.13 (2026-09-18)

Full Changelog: v1.2.12...v1.2.13

Features

deployments: add accept_shapeless_risk opt-out on deployments.create for intentionally creating a deployment without a pre-validated shape; document the deployment_shape="default" sentinel that asks the server to pick a default validated shape
training-sdk: add accept_shapeless_risk to DeploymentConfig for the manual shapeless creation path, rejected client-side when combined with deployment_shape

Bug Fixes

training-sdk: drop redundant TITO metric series (turn/model_tokens, turn/requested_output_tokens, trajectory/model_tokens_processed, agent/wall_seconds, and the debug/* artifact-write family) and remove the observe_agent_wall API; trajectory summaries and Harbor trial phase brackets carry the same timing information

Open source →

1.2.1216 Sept 2026Deprecated2 sources agree

training-sdk: deprecate explicit reference shapes for LoRA policies in favor of shared-session base references

1.2.12 (2026-09-15)

Full Changelog: v1.2.11...v1.2.12

Bug Fixes

training-sdk: stop sending the obsolete managed-rollout deployment annotation
training-sdk: wait for control-plane readiness when reattaching rollout deployments
training-sdk: avoid copying dense tensor values when sizing requests for every client
training-sdk: automatically negotiate optimized training requests for dedicated and serverless trainers advertising comms="v2"; missing or unknown capabilities preserve comms="v1" and its original request format without a manual flag
training-sdk: deprecate explicit reference shapes for LoRA policies in favor of shared-session base references

Open source →

1.2.1131 Aug 20262 sources agree

training-sdk: add environment-local TITO sidecars with full-history prompt construction by default and an experimental incremental renderer protocol

1.2.11 (2026-08-31)

Full Changelog: v1.2.10...v1.2.11

Features

training-sdk: add environment-local TITO sidecars with full-history prompt construction by default and an experimental incremental renderer protocol
training-sdk: preserve and attribute completion serving metrics when merging response headers with final performance metrics
training: support explicit reservation targeting for dedicated and managed training jobs

Open source →

1.2.1030 Aug 20262 sources agree

training-sdk: expose merged-base output precision with source-format export as the default

1.2.10 (2026-08-30)

Full Changelog: v1.2.9...v1.2.10

Features

training-sdk: expose merged-base output precision with source-format export as the default

Bug Fixes

training-sdk: fail trainer readiness waits fast when the job stops instead of polling to the readiness timeout
training-sdk: preserve canonical structured status details for managed recipe failures
training-sdk: preserve Lifecycle, Tinker, and serverless gateway error classifications

Open source →

1.2.912 Aug 20262 sources agree

training-sdk: preserve per-request affinity for serverless training sampling

1.2.9 (2026-08-12)

Full Changelog: v1.2.8...v1.2.9

Bug Fixes

training-sdk: preserve per-request affinity for serverless training sampling

Open source →

1.2.812 Aug 20262 sources agree

training: default managed SFT/DPO and dedicated Training API trainers to reservation-first placement with an explicit opt-out

1.2.8 (2026-08-12)

Full Changelog: v1.2.7...v1.2.8

Features

training: default managed SFT/DPO and dedicated Training API trainers to reservation-first placement with an explicit opt-out

Open source →

1.2.712 Aug 20262 sources agree

training: support reservation-first placement for managed SFT/DPO and dedicated Training API trainers

1.2.7 (2026-08-07)

Full Changelog: v1.2.6...v1.2.7

Features

training: support reservation-first placement for managed SFT/DPO and dedicated Training API trainers
training-sdk: expose managed SFT and DPO renderer Hugging Face repo ID selection

Open source →

1.2.63 Aug 20262 sources agree

training: support CMEK-encrypted managed RFT sampler hot-loads and isolate reference trainers

1.2.6 (2026-08-03)

Full Changelog: v1.2.5...v1.2.6

Features

training: support CMEK-encrypted managed RFT sampler hot-loads and isolate reference trainers
training-sdk: support repeated per-adapter model creation on one managed trainer

Bug Fixes

release: avoid false promotion-source failures after successful SDK releases

Open source →

1.2.531 Jul 20262 sources agree

training-sdk: log backend training shape override warnings

1.2.5 (2026-07-31)

Full Changelog: v1.2.4...v1.2.5

Bug Fixes

training-sdk: log backend training shape override warnings

Open source →

1.2.427 Jul 20262 sources agree

training-sdk: expose MoE model capability lookup for Router Replay

1.2.4 (2026-07-27)

Full Changelog: v1.2.3...v1.2.4

Features

training-sdk: expose MoE model capability lookup for Router Replay
training-sdk: expose structured Training API errors
training-sdk: propagate training skill attribution headers

Bug Fixes

training-sdk: gate RLOR inactivity cleanup compatibility to Fireworks-owned jobs

Open source →

1.2.321 Jul 20262 sources agree

require generated promotion provenance before public SDK releases

1.2.3 (2026-07-21)

Full Changelog: v1.2.2...v1.2.3

CI

require generated promotion provenance before public SDK releases

Open source →

1.2.221 Jul 20262 sources agree

1.2.2 (2026-07-21) Full Changelog: v1.2.1...v1.2.2

1.2.2 (2026-07-21)

Full Changelog: v1.2.1...v1.2.2

Open source →

1.2.121 Jul 20262 sources agree

serverless-rl: single retry owner + structured sampling error (SDK)

1.2.1 (2026-07-20)

Full Changelog: v1.2.0...v1.2.1

Features

serverless-rl: single retry owner + structured sampling error (SDK) (7113d1a)
training-sdk: adjust rollout concurrency within RL steps (68595ff)
training-sdk: expose routing matrices in sample responses (81d52c2)
training: serverless cross-run resume — SDK config-from-checkpoint + cookbook unlock (PR3+PR5) (a284467)

Bug Fixes

training-sdk: support Tinker 0.23 and unblock 1.2.1 release (7415ed8)
training: pin compatible pyqwest transport (53ae2be)

Refactors

training-sdk: add sampling concurrency protocol (9572c14)
async-rl: use a batch-native rollout producer (04ff3ea)

Open source →

1.2.011 Jul 20262 sources agree

## 1.2.0 (2026-07-11) Full Changelog: v1.2.0-alpha.89...v1.2.0

1.2.0 (2026-07-11)

Full Changelog: v1.2.0-alpha.89...v1.2.0

Open source →

1.2.0a8910 Jul 2026pre-release2 sources agree

training-sdk: sanitize public package contents

1.2.0-alpha.89 (2026-07-10)

Full Changelog: v1.2.0-alpha.88...v1.2.0-alpha.89

Bug Fixes

training-sdk: sanitize public package contents (f0944c6)

Chores

training-sdk: raise async httpx connection limits (a562db0)

Open source →

1.2.0a889 Jul 2026pre-release2 sources agree

training-sdk: default sampler concurrency to adaptive

1.2.0-alpha.88 (2026-07-09)

Full Changelog: v1.2.0-alpha.87...v1.2.0-alpha.88

Features

training-sdk: default sampler concurrency to adaptive (53deb8e)

Open source →

1.2.0a876 Jul 2026pre-release2 sources agree

training-sdk: declare urllib3 runtime dependency

1.2.0-alpha.87 (2026-07-06)

Full Changelog: v1.2.0-alpha.86...v1.2.0-alpha.87

Bug Fixes

training-sdk: declare urllib3 runtime dependency (8c2c9d5)
training: return 403 for training quota errors and surface the real cause (579b823)

Open source →

1.2.0a865 Jul 2026pre-release2 sources agree

managed: keep auto-shape selection when trainer.extra_args is set

1.2.0-alpha.86 (2026-07-05)

Full Changelog: v1.2.0-alpha.85...v1.2.0-alpha.86

Bug Fixes

managed: keep auto-shape selection when trainer.extra_args is set (a4b2a72)
training-sdk: reject tombstone trainer reattach on create/poll (3c5ceee)

Open source →

1.2.0a8526 Jun 2026pre-release2 sources agree

training: propagate auto-shape max context metadata

1.2.0-alpha.85 (2026-06-26)

Full Changelog: v1.2.0-alpha.84...v1.2.0-alpha.85

Bug Fixes

training: propagate auto-shape max context metadata (15d0a4d)

Open source →

1.2.0a8425 Jun 2026pre-release2 sources agree

training-sdk: surface serverless session (on the service) + run (on the training client)

1.2.0-alpha.84 (2026-06-24)

Full Changelog: v1.2.0-alpha.83...v1.2.0-alpha.84

Features

training-sdk: surface serverless session (on the service) + run (on the training client) (ad77c33)

Bug Fixes

sdk: guard training deployment shapes (4b3ebbb)
training: list and promote serverless session checkpoints by run (aefd3a8)
training: require LoRA reference shapes (88a96e3)

Open source →

1.2.0a8318 Jun 2026pre-release2 sources agree

public-repos: make SDK pyproject the version source

1.2.0-alpha.83 (2026-06-18)

Full Changelog: v1.2.0-alpha.82...v1.2.0-alpha.83

Chores

public-repos: make SDK pyproject the version source (c349c92)

Open source →

1.2.0a8216 Jun 2026pre-release2 sources agree

cookbook/rl: expose lora_alpha in rl_loop and managed provisioning

1.2.0-alpha.82 (2026-06-16)

Full Changelog: v1.2.0-alpha.81...v1.2.0-alpha.82

Features

cookbook/rl: expose lora_alpha in rl_loop and managed provisioning (5a67b14)

Open source →

1.2.0a8115 Jun 2026pre-release2 sources agree

training-sdk: full-distribution RL rollouts + default LoRA alpha=32

1.2.0-alpha.81 (2026-06-13)

Full Changelog: v1.2.0-alpha.80...v1.2.0-alpha.81

Bug Fixes

training-sdk: full-distribution RL rollouts + default LoRA alpha=32 (8e3df52)

Open source →

1.2.0a8012 Jun 2026pre-release2 sources agree

rlor: merged-base LoRA save + reject ineffective warm_start_from LoRA

1.2.0-alpha.80 (2026-06-12)

Full Changelog: v1.2.0-alpha.79...v1.2.0-alpha.80

Features

rlor: merged-base LoRA save + reject ineffective warm_start_from LoRA (#28107) (8b73705)

Bug Fixes

training-sdk: report probed deployments as ready (#29687) (d766124)

Open source →

1.2.0a7910 Jun 2026pre-release2 sources agree

## 1.2.0-alpha.79 (2026-06-09) Full Changelog: v1.2.0-alpha.78...v1.2.0-alpha.79

1.2.0-alpha.79 (2026-06-09)

Full Changelog: v1.2.0-alpha.78...v1.2.0-alpha.79

Open source →

1.2.0a789 Jun 2026pre-release2 sources agree

rlor: centralize rollout cleanup mode

1.2.0-alpha.78 (2026-06-08)

Full Changelog: v1.2.0-alpha.77...v1.2.0-alpha.78

Bug Fixes

rlor: centralize rollout cleanup mode (#29102) (d34d224)

Open source →

1.2.0a778 Jun 2026pre-release2 sources agree

serverless: sft_loop serverless mode + SDK session-checkpoint promote

1.2.0-alpha.77 (2026-06-06)

Full Changelog: v1.2.0-alpha.76...v1.2.0-alpha.77

Features

serverless: sft_loop serverless mode + SDK session-checkpoint promote (#28596) (470c845)

Open source →

1.2.0a764 Jun 2026pre-release2 sources agree

## 1.2.0-alpha.76 (2026-06-04) Full Changelog: v1.2.0-alpha.75...v1.2.0-alpha.76

1.2.0-alpha.76 (2026-06-04)

Full Changelog: v1.2.0-alpha.75...v1.2.0-alpha.76

Open source →

1.2.0a753 Jun 2026pre-release2 sources agree

## 1.2.0-alpha.75 (2026-06-03) Full Changelog: v1.2.0-alpha.74...v1.2.0-alpha.75

1.2.0-alpha.75 (2026-06-03)

Full Changelog: v1.2.0-alpha.74...v1.2.0-alpha.75

Open source →

1.2.0a742 Jun 2026pre-release2 sources agree

## 1.2.0-alpha.74 (2026-06-01) Full Changelog: v1.2.0-alpha.73...v1.2.0-alpha.74

1.2.0-alpha.74 (2026-06-01)

Full Changelog: v1.2.0-alpha.73...v1.2.0-alpha.74

Open source →

1.2.0a7322 May 2026pre-release

autorelease single source of truth

1.2.0-alpha.73 (2026-05-22)

Full Changelog: v1.2.0-alpha.72...v1.2.0-alpha.73

Chores

autorelease single source of truth (#90) (fc3f8c7)

Open source →

Additional notes

Full Changelog: v1.2.0-alpha.73...v1.2.0-alpha.73

Bug Fixes

training-sdk: keep forward-only reference trainer single-replica under HSDP (24eef8e)

Open source →

1.2.0a7221 May 2026pre-release2 sources agree

bootstrap public SDK release workflows

1.2.0-alpha.72 (2026-05-21)

Full Changelog: v1.2.0-alpha.71...v1.2.0-alpha.72

Chores

bootstrap public SDK release workflows (#83) (7dcb03e)
bootstrap promotion pipeline workflows (#86) (e0042ab)

Open source →

1.2.0a7119 May 2026pre-release

Full Changelog: v1.2.0-alpha.70...v1.2.0-alpha.71

Open source →

1.2.0a7015 May 2026pre-release2 sources agree

## 1.2.0-alpha.70 (2026-05-15) Full Changelog: v1.2.0-alpha.69...v1.2.0-alpha.70

1.2.0-alpha.70 (2026-05-15)

Full Changelog: v1.2.0-alpha.69...v1.2.0-alpha.70

Open source →

1.2.0a697 May 2026pre-release2 sources agree

sdk/training/hotload: forward snapshot URI via x-fireworks-hot-load-source-url header

1.2.0-alpha.69 (2026-05-06)

Full Changelog: v1.2.0-alpha.68...v1.2.0-alpha.69

Features

sdk/training/hotload: forward snapshot URI via x-fireworks-hot-load-source-url header (a842b84)

Open source →

1.2.0a686 May 2026pre-release2 sources agree

## 1.2.0-alpha.68 (2026-05-06) Full Changelog: v1.2.0-alpha.67...v1.2.0-alpha.68

1.2.0-alpha.68 (2026-05-06)

Full Changelog: v1.2.0-alpha.67...v1.2.0-alpha.68

Open source →

1.2.0a675 May 2026pre-release2 sources agree

## 1.2.0-alpha.67 (2026-05-05) Full Changelog: v1.2.0-alpha.66...v1.2.0-alpha.67

1.2.0-alpha.67 (2026-05-05)

Full Changelog: v1.2.0-alpha.66...v1.2.0-alpha.67

Open source →

1.2.0a665 May 2026pre-release2 sources agree

restore prerelease versioning in release-please-config.json

1.2.0-alpha.66 (2026-05-01)

Full Changelog: v1.2.0-alpha.65...v1.2.0-alpha.66

Reverts

restore prerelease versioning in release-please-config.json (#132) (a574884)

Open source →

1.2.0a651 May 2026pre-release2 sources agree

sdk/deployment: SSE-truncation retry hardening

1.2.0-alpha.65 (2026-05-01)

Full Changelog: v1.1.0-alpha.64...v1.2.0-alpha.65

Features

sdk/deployment: SSE-truncation retry hardening (#129) (fcfdba6)

Open source →

1.1.0a6428 Apr 2026pre-release2 sources agree

promote: accept 4-segment name= kwarg

1.1.0-alpha.64 (2026-04-28)

Full Changelog: v1.0.1-alpha.63...v1.1.0-alpha.64

Features

promote: accept 4-segment name= kwarg (#126) (7ecd22f)
support setting headers via env (59ba573)

Bug Fixes

use correct field name format for multipart file arrays (8aa4a63)

Chores

internal: more robust bootstrap script (b535ded)

Open source →

1.0.1a6322 Apr 2026pre-release2 sources agree

## 1.0.1-alpha.63 (2026-04-22) Full Changelog: v1.0.0-alpha.63...v1.0.1-alpha.63

1.0.1-alpha.63 (2026-04-22)

Full Changelog: v1.0.0-alpha.63...v1.0.1-alpha.63

Open source →

1.0.0a6322 Apr 2026pre-release2 sources agree

downstream compat shims, CI gates, and test coverage

1.0.0-alpha.63 (2026-04-22)

Full Changelog: v1.0.0-alpha.62...v1.0.0-alpha.63

Features

downstream compat shims, CI gates, and test coverage (#62) (2f59a2f)

Bug Fixes

compat: make acreate(stream=True) iterable without await (#123) (a8e3121)

Performance Improvements

client: optimize file structure copying in multipart requests (da87cc2)

Documentation

rollouts: add BYOT RL rollout guides (4b9fe46)

Open source →

1.0.0a6216 Apr 2026pre-release2 sources agree

ensure file data are only sent as 1 parameter

1.0.0-alpha.62 (2026-04-16)

Full Changelog: v1.0.0-alpha.61...v1.0.0-alpha.62

Bug Fixes

ensure file data are only sent as 1 parameter (c7fde6c)

Open source →

1.0.0a619 Apr 2026pre-release2 sources agree

sdk: deployment update, delta-chain reset, and stale-error handling for re-attach

1.0.0-alpha.61 (2026-04-09)

Full Changelog: v1.0.0-alpha.60...v1.0.0-alpha.61

Features

sdk: deployment update, delta-chain reset, and stale-error handling for re-attach (#116) (6d662aa)
training: add purpose and managed_by fields to TrainerJobConfig (8b7790c)

Bug Fixes

client: preserve hardcoded query params when merging with user params (fc43cb4)

Documentation

training: clarify purpose and managed_by are managed-service internal fields (4e06d9e)
training: make purpose and managed_by docstrings deliberately terse (defb83d)

Open source →

1.0.0a606 Apr 2026pre-release2 sources agree

training-sdk: add create_base_training_client for LoRA self-reference

1.0.0-alpha.60 (2026-04-06)

Full Changelog: v1.0.0-alpha.59...v1.0.0-alpha.60

Features

training-sdk: add create_base_training_client for LoRA self-reference (f3fef7c)

Open source →

1.0.0a593 Apr 2026pre-releaseDeprecated2 sources agree

deprecate hot_load_deployment_id with runtime warning

1.0.0-alpha.59 (2026-04-03)

Full Changelog: v1.0.0-alpha.58...v1.0.0-alpha.59

Features

deprecate hot_load_deployment_id with runtime warning (#105) (e430226)

Open source →

1.0.0a583 Apr 2026pre-release2 sources agree

training-sdk: disable delta hotload for LoRA adapters

1.0.0-alpha.58 (2026-04-03)

Full Changelog: v1.0.0-alpha.57...v1.0.0-alpha.58

Bug Fixes

training-sdk: disable delta hotload for LoRA adapters (#114) (4cfd1e7)

Open source →

1.0.0a573 Apr 2026pre-release2 sources agree

## 1.0.0-alpha.57 (2026-04-03) Full Changelog: v1.0.0-alpha.56...v1.0.0-alpha.57

1.0.0-alpha.57 (2026-04-03)

Full Changelog: v1.0.0-alpha.56...v1.0.0-alpha.57

Open source →

1.0.0a563 Apr 2026pre-release2 sources agree

add skip_validations to TrainerJobConfig

1.0.0-alpha.56 (2026-04-03)

Full Changelog: v1.0.0-alpha.55...v1.0.0-alpha.56

Features

add skip_validations to TrainerJobConfig (#111) (c0498a5)

Open source →

1.0.0a552 Apr 2026pre-release2 sources agree

support versioned paths in resolve_training_profile

1.0.0-alpha.55 (2026-04-02)

Full Changelog: v1.0.0-alpha.54...v1.0.0-alpha.55

Features

support versioned paths in resolve_training_profile (#107) (ae77373)

Open source →

1.0.0a542 Apr 2026pre-release

training-sdk: close dogfooding gaps for local loops

Bug Fixes

training-sdk: close dogfooding gaps for local loops (#103)

Open source →

Additional notes

Full Changelog: v1.0.0-alpha.53...v1.0.0-alpha.54

Bug Fixes

training-sdk: close dogfooding gaps for local loops (#103) (0907345)

Open source →

1.0.0a531 Apr 2026pre-release2 sources agree

deployment_shape property returns versioned path

1.0.0-alpha.53 (2026-04-01)

Full Changelog: v1.0.0-alpha.52...v1.0.0-alpha.53

Bug Fixes

deployment_shape property returns versioned path (#102) (74c2c61)

Open source →

1.0.0a5230 Mar 2026pre-release2 sources agree

add hot_load_trainer_job to DeploymentConfig

1.0.0-alpha.52 (2026-03-30)

Full Changelog: v1.0.0-alpha.51...v1.0.0-alpha.52

Features

add hot_load_trainer_job to DeploymentConfig (#89) (51d75a3)
use account-level checkpoint promotion endpoint (#88) (cd9efe9)
use account-level checkpoint promotion endpoint (#88) (f19ae68)

Bug Fixes

WeightSyncer.save_dcp passes invalid timeout kwarg to save_state (#100) (5a1a386)

Open source →

1.0.0a5127 Mar 2026pre-release2 sources agree

restore missing urlencode import in trainer.py

1.0.0-alpha.51 (2026-03-27)

Full Changelog: v1.0.0-alpha.50...v1.0.0-alpha.51

Bug Fixes

restore missing urlencode import in trainer.py (#93) (a5b08ba)

Open source →

Your coding agent can read these notes before it upgrades. Set up the MCP server →

fireworks-ai

Release timeline

Releases

1.2.19 (2026-10-02)

Features

1.2.18 (2026-10-01)

Features

Bug Fixes

1.2.17 (2026-09-26)

Features

1.2.16 (2026-09-25)

Features

1.2.15 (2026-09-22)

Features

Bug Fixes

1.2.14 (2026-09-18)

Features

1.2.13 (2026-09-18)

Features

Bug Fixes

1.2.12 (2026-09-15)

Bug Fixes

1.2.11 (2026-08-31)

Features

1.2.10 (2026-08-30)

Features

Bug Fixes

1.2.9 (2026-08-12)

Bug Fixes

1.2.8 (2026-08-12)

Features

1.2.7 (2026-08-07)

Features

1.2.6 (2026-08-03)

Features

Bug Fixes

1.2.5 (2026-07-31)

Bug Fixes

1.2.4 (2026-07-27)

Features

Bug Fixes

1.2.3 (2026-07-21)

CI

1.2.2 (2026-07-21)

1.2.1 (2026-07-20)

Features

Bug Fixes

Refactors

1.2.0 (2026-07-11)

1.2.0-alpha.89 (2026-07-10)

Bug Fixes

Chores

1.2.0-alpha.88 (2026-07-09)

Features

1.2.0-alpha.87 (2026-07-06)

Bug Fixes

1.2.0-alpha.86 (2026-07-05)

Bug Fixes

1.2.0-alpha.85 (2026-06-26)

Bug Fixes

1.2.0-alpha.84 (2026-06-24)

Features

Bug Fixes

1.2.0-alpha.83 (2026-06-18)

Chores

1.2.0-alpha.82 (2026-06-16)

Features

1.2.0-alpha.81 (2026-06-13)

Bug Fixes

1.2.0-alpha.80 (2026-06-12)

Features

Bug Fixes

1.2.0-alpha.79 (2026-06-09)

1.2.0-alpha.78 (2026-06-08)

Bug Fixes

1.2.0-alpha.77 (2026-06-06)

Features

1.2.0-alpha.76 (2026-06-04)

1.2.0-alpha.75 (2026-06-03)

1.2.0-alpha.74 (2026-06-01)