Together AI
@togethercompute
Followers
52K
Following
2K
Media
697
Statuses
2K
Accelerate training, fine-turning, and inference on performance-optimized GPU clusters.
San Francisco, CA
Joined November 2022
The RC period runs for ~1 month, and we’re actively looking for feedback before GA. We’ve also published a detailed Python SDK Migration Guide covering breaking changes, API-by-API notes, and troubleshooting tips. 👉Read the announcement + migration guide in our latest blog:
together.ai
0
0
2
Why v2? 👉Faster & more maintainable 👉Strong typing + better DX 👉New APIs and uv support 👉New exception hierarchy
1
0
2
We’re excited to announce the Together Python SDK v2.0 Release Candidate 🐍 This major upgrade is a type-safe, OpenAPI-generated client designed to closely mirror our API, improve performance, and make it much easier to ship new features going forward.
1
0
7
Excited about our partnership with @nvidia, and how we are jointly helping deliver large scale inference for leading AI-natives through custom optimizations.
⚡ @togethercompute deployed NVIDIA GB200 NVL72 systems to deliver large-scale inference workloads on frontier MoE models, such as DeepSeek-V3. 🔗 Learn more about how we're overcoming MoE scaling bottlenecks with extreme codesign: https://t.co/bKXrHrSmMa
1
3
10
🎲New demo: Train LLMs to play BlackJack using GRPO on Kubernetes! Shows distributed RL training with OpenEnv + TorchForge on Together Instant Cluster. Qwen learns optimal strategy through reinforcement learning with group-relative policy optimization. Check it out:
2
1
14
Congrats to @yutori_ai for the public launch of Scouts, their always-on agents that monitor the web and reliably handle everyday digital tasks. Always exciting to see AI-native companies like Yutori building on Together AI Inference and powering these agents at production scale.
Today, we're making Scouts available to everyone! Earlier this year, Scouts was born out of a simple observation — that so many of life's background (or even foreground!) tasks have a recurring flavor, e.g. house hunting, early stages of travel planning, sourcing leads,
0
3
11
Announcing LlamaCoder v3 – generate React apps in 1 prompt! • Multi-file generation for better apps • Monaco editor to view & export code • New models: GLM 4.6, Kimi K2, & Qwen 3 Coder 100% free, open source, and powered by @togethercompute.
26
73
611
We're excited to partner with @ServiceNow to bring Apriel-1.6-15B-Thinker to Together AI. Try it now!
together.ai
15B multimodal reasoning model with 131K context, scoring 57 on AA index, matching models 15x its size.
0
0
4
Highlights: 👉 On par with Qwen 235B A22B & DeepSeek v3.2 Exp on Artificial Analysis 👉 Fits on a single GPU for cost-efficient deployment 👉 Strong reasoning: AIME'25: 88, GPQA: 73, LCB: 81 👉 Enterprise ready: IFBench 69, Tau2 Telecom 69 👉 30% more token efficient than
1
0
1
Introducing @ServiceNow AI’s Apriel-1.6-15B-Thinker, a 15B multimodal reasoning model that matches the performance of 235B models while being 15x smaller. AI natives can now use Apriel-1.6-15B-Thinker on Together AI — and benefit from reliable inference at production scale.
1
3
20
And congratulations to the Together AI team members @realDanFu, @m_ryabinin, @ShangZhu18, @ben_athi, @_junxiong_wang for showcasing their latest research and pushing the boundaries of AI innovation. #NeurIPS2025 same time next year?
0
2
4
What an incredible time at @NeurIPSConf! 🧠 Huge thank you to the researchers and builders who visited our booth to deep dive into the AI Native Cloud and to those who joined us for exclusive events to network and connect with their peers – shout out to our co-hosts @NVIDIA,
2
2
5
It’s here. AI Agents Assemble, the biggest WeMakeDevs hackathon so far. $20,000 (₹18 lakh) in cash prizes 💰 Winners get job interview opportunities. Everyone who participates gets Google Summer of Code mentorship ☀️ First place alone wins $10,000 cash 🥇 Powered by some of
7
9
121
The OSS coding model that @ashVaswani and team at @essential_ai have been working on over the last year is now out and available to use on @togethercompute APIs and playground. It's particularly exciting that @essential_ai is doing fundamental work in pre-training and post
together.ai
8B instruction-tuned model with 32K context, exceptional agentic coding, advanced tool use, elite math & STEM reasoning
6
21
123
@essential_ai Try Rnj-1 Instruct now on Together AI
together.ai
8B instruction-tuned model with 32K context, exceptional agentic coding, advanced tool use, elite math & STEM reasoning
0
3
26
@essential_ai Highlights: 👉 Dominant on SWE-bench, BFCL, and Enamel 👉 Strong tool use and function calling 👉 32K context window 👉 Apache 2.0 licensed, fully open weights 👉 Trained on 8.4T tokens with Muon optimizer
1
3
23
Introducing Rnj-1 Instruct from @essential_ai, an open-source 8B model engineered for agentic coding and STEM tasks. AI natives can now use Rnj-1 Instruct on Together AI and benefit from reliable inference for production-scale software engineering and scientific workflows.
3
10
38
Excited to partner with @scaledcognition. Their new research achieves 70x faster training on hierarchical datasets — turning weeks of GPU time into hours. AI-native companies like Scaled Cognition push the boundaries of AI systems research on Together AI.
New blog post - Prompt Trees: Training-time Prefix Caching. By the research team at @scaledcognition. TL;DR: Training speedups of up to 70x on tree-structured data. Not 70%. _70x_. https://t.co/EYD96dHAHk (preprint version coming soon)
2
1
8
We’re taking the first step toward production-grade RL on the AI Native Cloud. Together AI + @AIatMeta's team are partnering to bring high-performance reinforcement learning to real agentic systems — long-horizon reasoning, tool use, and multi-step workflows. Check out the
together.ai
Together AI and PyTorch partner to deliver open-source reinforcement learning for agentic AI systems. Build, train, and deploy advanced AI agents with integrated RL on the Together platform.
3
2
22