Transformers Release v5.14.0

UpdateJul 15, 2026VerifiedAdded Sep 22, 2026

Release v5.14.0 New Model additions Inkling (fresh from Thinking Machines): 975B total, 41B active Add Inkling model #47347 by @molbap @Cyrilvallez @eustlb and @zucchini-nlp Inkling is a general-purpose multimodal model that accepts text, image and audio inputs and generates text outputs.

AI agents Vector search Streaming

Read Hugging Face's release notes

https://github.com/huggingface/transformers/releases/tag/v5.14.0

Summaries of vendors' own notes. Product names and logos belong to their owners; logos via logo.dev.

More Transformers releases

TransformersJul 16, 2026

Transformers Patch release: v5.14.1

Patch release v5.14.1 This patch solves a few issues which appeared when integrating Inkling model, most notably an issue affecting models using EncoderDecoderCache during assisted generation. It also fixes an issue that could appear during prefill with StaticCache and sdpa without padding for Inkling which uses a position_bias.

UpdateSource

TransformersJul 11, 2026

Transformers Patch release v5.13.1

Patch release v5.13.1 This patch is focused on enabling transformers for the latest release of vllm! Be more defensive with remap_legacy_layer_types for custom models ( #47245 ) from @hmellor Fix custom code which doesn't know about the new linear layer type names ( #47174 ) from @hmellor Fix case where _LazyAutoMapping.register is passed a str key ( #47148...

UpdateSource

TransformersJul 3, 2026

Transformers Release v5.13.0

Release v5.13.0 New Model additions KimiK 2.5, 2.6, and 2.7 This release includes the architecture for Kimi 2.5 which is used by 2.5-2.7: Kimi K2.5 is an open-source, native multimodal agentic model that advances practical capabilities in long-horizon coding, coding-driven design, proactive autonomous execution, and swarm-based task orchestration.

UpdateAI agentsPerformanceSource

TransformersAug 10, 2026

Transformers Release: v5.15.0

Release v5.15.0 New Model additions Meta Muse Glimmer Muse Glimmer, released today, is Meta’s new multimodal model, especially designed for agentic use cases. Distilled from Muse to 30B parameters, and released under the Apache 2.0 license, it can be deployed to local setups for privacy-aware applications such as coding, document analysis, personal...

UpdateAI agentsObservabilityDeveloper toolsSource

TransformersAug 19, 2026

Transformers Patch release: v5.15.1

Patch release v5.15.1 This patch most notably solves a few issues with DFlash and MTP candidate generators, as well as an issue where images could sometimes not be processed on accelerator if using Lanczos filter.

UpdateSource

TransformersAug 26, 2026

Transformers Release v5.16.1

Release v5.16.1 This is a special release as we include GLM! (and a few small fixes) GLM-5.3-Flash GLM-5.3-Flash, the first natively multimodal model in the GLM-5 series.

UpdateAI agentsLLMsSecuritySource

TransformersAug 26, 2026

Transformers Release: v5.16.0

Release v5.16.0 New Model additions Qwen4-Exp Qwen4-Exp builds on Qwen3.5's hybrid text and multimodal architecture with three key components: GatedResidual (GR), Qwen Sparse Attention (QSA), and Per-Layer Embedding (PLE). GR is a Qwen-developed residual architecture that combines Hyper-Connection with GatedNorm.

UpdateVector searchSQLStreamingSource

TransformersSep 10, 2026

Transformers Release 5.17.0

Release v5.17.0 New Model additions HYV4 Hy4-Preview is a 780B-parameter mixture-of-experts language model that activates 49B parameters per token. Each MoE layer holds 256 routed experts plus one always-active shared expert and routes every token to 8 of them. The context window is 1M tokens.

PreviewSQLStreamingSource

Also shipped on Jul 15, 2026

Google Cloud BigQueryJul 15, 2026

BigQuery supports data governance tags , which let you enforce column-level security and data masking

BigQuery supports data governance tags , which let you enforce column-level security and data masking. Data governance tags are a type of Resource Manager tag that you can attach to sensitive columns and use in BigQuery data policies to grant conditional access to your users. This feature is in Preview .

PreviewSQLGovernanceSecuritySource

Snowflake Snowflake Native Apps and MarketplaceJul 15, 2026

Integrated SaaS listings on Snowflake Marketplace (General availability)

Select Integrated SaaS listings now let you start a free trial of a provider’s application directly from a Marketplace listing. This experience replaces the trials previously offered through Snowflake Partner Connect and brings these trials into the Marketplace, where discovery, trial, and purchase all happen in one place.

GASource

Snowflake SnowflakeJul 15, 2026

Consumer-controlled maintenance policies , General Availability

Consumer-controlled maintenance policies for Snowflake Native Apps are now generally available. Consumers can define when Snowflake Native App upgrades happen in their accounts by creating a maintenance policy and applying it to an account or an individual app. For more information, see Consumer-controlled maintenance policies .

GAGovernanceDeveloper toolsSource

Databricks Databricks Data Intelligence PlatformJul 15, 2026

Thinking Machine Labs Inkling now available as a Databricks-hosted model in Public Preview

Model Serving now supports Thinking Machine Labs Inkling as a Databricks-hosted model. This mixture of experts (MoE) model supports text and image inputs, a 1 million token context window, and is optimized for coding, reasoning, and agentic tool-use tasks. You can access this model using Foundation Model APIs . Inkling is available in Public Preview .

PreviewAI agentsGovernanceSource

Databricks Databricks Data Intelligence PlatformJul 15, 2026

Custom URLs for your Databricks account (Public Preview)

You can now claim and enable a Custom URL that gives your Databricks account a single, branded entry point (for example, acme.databricks.com ) for all users and workspaces. See Give your Databricks account a custom URL .

PreviewSource

Databricks Databricks Data Intelligence PlatformJul 15, 2026

AI Search High QPS is now generally available

AI Search High QPS is now generally available for standard endpoints. Set a target QPS on standard AI Search endpoints to power high-throughput, real-time workloads like search bars, recommendation systems, and entity matching. High QPS is enabled by default, but existing endpoints are not affected.

GAStreamingPricingPerformanceSource