Releasesby GlyphPost-training: GLM-5.3 Flash
zai-org/GLM-5.3-Flash is available for post-training : LoRA training on B200-SXM , a 159,744-token training context, a 163,840-token generator context, rank up to 32, and up to five adapter sessions per resource. The cookbook renders its prompts with the GLM codec.
Also shipped Nov 2, 2026Together AI in November 2026 →
Managed warehouse for MLflow (Beta)
Managed warehouse for MLflow is available in Beta, enabling users to view Unity Catalog traces in the MLflow UI without manually choosing a SQL warehouse.
Check out GitHub code without long-lived credentials on self-hosted agents
Self-hosted agents can now use temporary, read-only GitHub App tokens for repository checkout. This eliminates the need to distribute and rotate long-lived SSH keys or personal access tokens across your agent fleet.
Serving tuned Gemini 3.x models on global endpoints
Serving tuned Gemini 3.x models on global endpoints is available in General Availability (GA) . You can copy fine-tuned Gemini 3.x models to the global region using the Model Registry models:copy API, deploy them to global serving endpoints, and send inference requests directly to aiplatform.googleapis.com .
[Launched] Generally Available: Change from virtual network integration to a private endpoint capable network configuration in Azure Database for PostgreSQL flexible server
You now have greater flexibility in network design for Azure Database for PostgreSQL Flexible Server workloads. Now you can change existing VNet-integrated servers from delegated-subnet networking to a private endpoint-capable configuration, enabling secu
[Launched] Generally Available: Restart databases in Azure SQL DB via the Azure portal
You can now restart or fail over databases directly from the Azure portal, including single databases, Hyperscale databases, and databases in elastic pools, without writing scripts against the Rest API or contacting support. Learn more.
Lakehouse caching for cross-cloud data access is generally available (GA)
Lakehouse caching for cross-cloud data access is generally available (GA). Cross-cloud caching is disabled by default. You can opt in to caching on a Lakehouse federated catalog or a BigQuery cross-cloud connection to cache remote files in your Google Cloud region. This helps reduce cross-cloud egress costs and accelerate subsequent queries.
More from Together AI platformEvery Together AI platform release →
Messages API and Responses API
Together AI now supports the Anthropic Messages API at POST /v1/messages and the OpenAI Responses API at POST /v1/responses . To run an app built on either API on Together, change its API key, base URL, and model.
Post-training: FP4 preparation for MiniMax M3.1 Flash
You can now deploy a MiniMax M3.1 Flash adapter on dedicated model inference by first preparing it for FP4 inference . The base model already ships in its serving format, so the job only merges the adapter, which takes about 1 hour.
Post-training: MiniMax M3.1 Flash
Together AI now supports LoRA training for the MiniMax M3.1 Flash model, featuring a 262,144-token context window.
New serverless models
MiniMax-M3.1-Flash is now available on the Together AI serverless platform, supporting a 1M context window, function calling, and structured outputs.
Together SDK and CLI 3.0
Together Python SDK 3.0, TypeScript SDK 1.0, and Together CLI 3.0 are now generally available.
What's new:
Together AI has moved dedicated model inference and GPU clusters out of beta, improved chat completion speeds, and added billing usage reporting to the SDK.
Applied credit types in billing usage
Billing usage line items now include an applied_credit object, identifying whether usage was covered by free, prepaid, or postpaid credits.
Model deprecations
Together AI has deprecated Llama-3.3-70B-Instruct-Turbo and DeepSeek-V4-Flash-0731 on its serverless platform. Users are advised to migrate to newer model versions.