Together AI platform Release notes from Together AI

Post-training: Kimi K3.1 Pro

Together AI now supports LoRA post-training for the Kimi K3.1 Pro model with a 163,840-token training context.

UpdateNov 4

Post-training: MiniMax M3 retired

MiniMaxAI/MiniMax-M3 is no longer available for new post-training model resources. Use MiniMaxAI/MiniMax-M3.1-Flash instead. Existing MiniMax M3 model resources keep running, and their checkpoints stay available for download. See Supported models .

DeprecationNov 3

5 releases →

Post-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.

UpdateNov 2

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.

UpdateOct 23

Together SDK and CLI 3.0

Together Python SDK 3.0, TypeScript SDK 1.0, and Together CLI 3.0 are now generally available.

GAOct 22

5 releases →

Longer post-training context for Kimi K3

moonshotai/Kimi-K3 on post-training now supports a 163,840-token training context (up from 65,536) and a 167,936-token generator context (up from 69,632). See Supported models .

UpdateOct 19

New dedicated endpoint models

The following models are now available for deployment on dedicated endpoints : deepseek-ai/DeepSeek-V4.1-Flash-1017 .

UpdateOct 18

Post-training: Muse Spark Open

meta-models/Muse-Spark-Open-816B-A42B 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 one adapter session per resource. See Supported models .

UpdateOct 17

Geo-specific inference

You can now keep serverless inference processing in the United States or Canada. Set your client's base URL to https://us-api.together.ai/v1 for the United States or https://ca-api.together.ai/v1 for Canada, using your existing API key and model IDs. There are no GPUs to provision.

UpdateOct 16

New serverless models

The following models are now available on serverless : meta-models/Muse-Spark-Open-816B-A42B : 1,048,576 context length, NVFP4 quantization, function calling, and structured outputs. Pricing: $1.25 input / $4.25 output / $0.15 cached input (per 1M tokens).

UpdateOct 15

What's included:

- Three training methods: Supervised fine-tuning (SFT) , direct preference optimization (DPO) , and reinforcement learning with group relative policy optimization (GRPO) . Each method trains a low-rank adapter (LoRA) or the full model weights. Custom loss functions : Train with any objective you compute in PyTorch from the model's log-probabilities.

BetaOct 13

Dynamic rate limits for serverless retired

Serverless no longer applies dynamic rate limits that scale with your recent traffic. Most users will no longer encounter rate limits, though requests can still be limited during periods of high demand. If you get a 429 Too Many Requests response, reduce your request rate and spread out bursts.

DeprecationOct 2

Together Link beta

Together Link runs six coding agents on models hosted by Together AI: Claude Code, Codex, OpenCode, and Pi Code in the terminal, plus Claude Desktop (including Cowork) and ChatGPT Desktop. Install it with one command, launch your agent through it, and your normal agent configuration stays untouched. It's now in beta on macOS and Linux.

BetaOct 1

4 releases →

Batch API files are now retained for 7 days

The input file you upload for a batch job, along with the output and error files the job produces, are now retained for 7 days. After that the files are no longer accessible, so download your results before the window closes. Reusing an uploaded input file across batch jobs also works only within that window. See Batch inference .

UpdateSep 29

TogetherLink beta

TogetherLink runs coding agents such as Claude Code and Codex, plus the Claude and ChatGPT desktop apps, on models hosted by Together AI. It's now in beta, with automatic model routing, cost tracking, and a headless mode.

BetaSep 24

New serverless models

The following models are now available on serverless : together/Tev1-4B-experimental : 32,768 context length. Pricing: $0.042 input / free output (per 1M tokens).

UpdateSep 23

Pricing update

The following models have lower pricing, effective September 22, 2026. All usage from that date forward is billed at the new rates (per 1M tokens): Qwen/Qwen3.7-Max : $2.50 → $1.50 (input), $7.50 → $4.50 (output). Qwen/Qwen3.8-Flash : $0.15 → $0.09 (input), $0.47 → $0.282 (output). See Serverless models for the full pricing catalog.

PricingSep 22

Automatic idle shutdown for dedicated deployments

Deployments can now stop themselves when they go unused. Set an inactivity timeout with --inactive-timeout (the inactiveTimeout field in the management API), and if the deployment serves no inference requests for that many minutes, it scales to zero replicas, releasing its hardware and stopping billing. See Automatic idle shutdown .

UpdateSep 16

Together CLI v2.34.0

Version 2.34.0 of the Together CLI and Python SDK carries the rollouts release above (the SDK surface is client.beta.endpoints.rollouts ) and also adds: HIPAA placement : tg beta endpoints deploy accepts --placement.hipaa to restrict a deployment to HIPAA-attested clusters. The Python SDK takes the same policy as compliance_policy on inline placement.

BetaSep 15

New serverless models

The following models are now available on serverless : MiniMaxAI/MiniMax-H3 : Pricing: $0.1391/sec at 2k.

UpdateSep 14

New serverless models

The following models are now available on serverless : deepseek-ai/DeepSeek-V4.1-Flash : 1,000,000 context length, FP8 quantization, function calling, and structured outputs. Pricing: $0.30 input / $1.20 output / $0.006 cached input (per 1M tokens).

UpdateSep 11

Preemptible compute for GPU clusters

Preemptible compute is now in public preview for Kubernetes GPU clusters. Alongside standard nodes, you can set a preemptible GPU target, and Together provisions toward it as spare capacity becomes available, at a flat discounted rate relative to on-demand.

PreviewSep 10

New models available for fine-tuning

You can now fine-tune the following models: zai-org/GLM-5.3 . See Supported models for the full list.

UpdateSep 8

Model deprecations

The following models are deprecated and will be removed from serverless on September 14, 2026: openai/gpt-oss-20b . Recommended replacement: Qwen/Qwen3.5-9B . google/gemma-4-31B-it . Recommended replacement: zai-org/GLM-5.3-Flash . thinkingmachines/Inkling-Small . Recommended replacement: zai-org/GLM-5.3-Flash . intfloat/multilingual-e5-large-instruct .

DeprecationSep 1

New serverless models

The following models are now available on serverless : Qwen/Qwen3.8-Flash : 1,000,000 context length. Pricing: $0.15 input / $0.47 output (per 1M tokens).

UpdateAug 31

active_sessions autoscaling metric for TTS

Dedicated endpoint autoscaling now accepts active_sessions , which targets concurrent open client WebSocket sessions per replica on a TTS deployment ( AVERAGE_VALUE only). The default inflight_requests metric also covers TTS handlers, so HTTP TTS traffic can keep scaling without a policy change. See Configure autoscaling .

UpdateAug 29

New serverless models

The following models are now available on serverless : zai-org/GLM-5.3 : 1,000,000 context length, FP4 quantization, function calling and structured outputs. Pricing: $1.40 input / $4.40 output / $0.26 cached input (per 1M tokens).

UpdateAug 28

Model deprecations

The following models have been deprecated and are no longer available on serverless: nvidia/Nemotron-3-ultra-550b-a55b . pearl-ai/gemma-4-31b-it . deepseek-ai/DeepSeek-V4-Pro . Use deepseek-ai/DeepSeek-V4-Pro-0813 instead. moonshotai/Kimi-K2.7-Code .

DeprecationAug 27

5 releases →

Batch jobs in the CLI

The Together CLI now includes a tg batches command group for batch inference : tg batches submit uploads a local JSONL file (or takes an existing file ID) and creates a job against chat.completions , audio.transcriptions , or audio.translations . tg batches list , retrieve , and cancel manage the job lifecycle, with ls and get as aliases.

UpdateAug 26

Model deprecations

The following models have been deprecated and are no longer available on serverless: google/gemma-3n-E4B-it . meta-llama/Llama-Guard-4-12B . These models are not supported by on-demand dedicated endpoints. See Deprecations for migration options.

DeprecationAug 25

Fine-tuning quality improvements

Training quality has improved for fine-tuning on the following models: Qwen/Qwen3.5-0.8B . Qwen/Qwen3.5-2B . Qwen/Qwen3.5-4B . Qwen/Qwen3.5-9B . Qwen/Qwen3.5-27B . Qwen/Qwen3.5-35B-A3B . Qwen/Qwen3.5-35B-A3B-Base . Qwen/Qwen3.5-122B-A10B . Qwen/Qwen3.5-397B-A17B . Qwen/Qwen3.6-27B . Qwen/Qwen3.6-35B-A3B . nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 .

UpdateAug 24