@togethercompute

Accelerate inference, model shaping, and pre-training on a research-optimized platform.

 San Francisco, CA
together.ai
Joined November 2022
  • Tweets3.2K
  • Following275
  • Followers64.3K
963Media

Together Post-Training: frontier-grade model shaping, the way that fits you

Article · @togethercompute
11
32
4
261
29,141
Together AI retweeted
Adit Jain
@aditjain198013h
Got to be a beta-tester for this! Pretty cool, agent-friendly and works for most-use cases! Custom Models ftw
Together AI
@togethercompute17h
Together Post-Training is live Turn open models into custom models you own, with SFT or RL. You define the data, rewards, environments and the training loop itself. We run the hard infrastructure underneath it, tuned for quality and speed.
1
2
5
1,012
Together Post-Training is live Turn open models into custom models you own, with SFT or RL. You define the data, rewards, environments and the training loop itself. We run the hard infrastructure underneath it, tuned for quality and speed.
16
24
12
139
24,512
more replies
Under the hood, every model gets a GPU setup we’ve already tested and rollout engines tuned for RL. We also keep training and generation in sync, so the quality you see in training holds up when you serve the model.
1
1
9
1,003
When you have a few good runs, compare the adapters side by side on one endpoint. Then deploy the best one on Together, or download the weights and run it anywhere. Available today on open models from Meta, Z.ai, Moonshot and Qwen, with more on the way. together.ai/post-training
2
7
816
deepseek v4.1 flash 1017 is now live on together ai deepseek trained this update with self-feedback rl to get better at building complete projects from scratch, from web apps and games to longer agentic coding tasks same price. try the model now ↓ together.ai/models/deepseek-…
6
5
1
53
3,698
Who is Space Bunny? 🐰🌕 Something's hopping onto Together AI 🔜⏳
7
10
2
347
64,502
Karl the Fog said no jets. the Blue Angels didn't get the memo ✈️ The Science of Speed, recap 👇🏻 @nvidia x @dtcp_capital
6
21
2,856
America’s next top model, Reflection’s Beam, is now live on Together AI. Trained on Together GPU Clusters, Beam is Reflection’s first open-weight model. • ~501B-parameter MoE, 23B active • Built for coding + long-horizon agentic work • 1M-token context window • Released under an Apache 2.0 release Try Beam now together.ai/models/reflectio…
16
8
2
74
6,308
Together AI retweeted
Vipul Ved Prakash
@vipulvedOct 15
Excited to be partnering with @AIatMeta to bring their new open models to @togethercompute. Meta Muse Spark Open now available! We are also releasing a free endpoint (with serious capacity) to allow our customers to test their workloads.
Together AI
@togethercomputeOct 15
Muse Spark Open is free on Together AI through Sunday at 12 PM PT After 12 PM PT on Oct 18, it routes to the paid endpoint at $1.25/M input + $4.25/M output. Try the free open weights model from Meta now: together.ai/models/muse-spar…
6
12
1
68
3,744
congrats to the StepFun team on releasing Step 5 Preview open weights today we’re excited to be a launch partner and to support another StepFun release as they bring Step 5 Preview to developers
StepFun
@StepFun_aiOct 15
Step 5 Preview is now open weights. BF16 and FP8 weights are live on Hugging Face and ModelScope, together with our technical blog. • Sparse MoE: 600B total, 27B active per token • 1M-token context window • Native text and vision input • Built for long-horizon agentic work: coding, software engineering, professional knowledge work and finance Frontier intelligence, open weights. Build what's next. Hugging Face: huggingface.co/stepfun-ai/St… ModelScope: modelscope.ai/models/stepfun… GitHub: github.com/stepfun-ai/Step-5…
5
2
25
3,011
Muse Spark Open is free on Together AI through Sunday at 12 PM PT After 12 PM PT on Oct 18, it routes to the paid endpoint at $1.25/M input + $4.25/M output. Try the free open weights model from Meta now: together.ai/models/muse-spar…
8
4
3
43
143,062
Muse Spark Open is now live on Together AI. We're proud to be a Day 0 launch partner for @AIatMeta's Muse Spark Open, the most capable US open-weights model to date, built for long-running agentic coding workflows that keep entire codebases in context.
7
2
6
36
6,311
more replies
It's agentic from the ground up. It browses the web, spawns and coordinates subagents, and uses computer-use capabilities across desktop, mobile, and browser. It states intent before acting, so you always know what it's doing and why.
1
585
Together Serverless Inference gives developers a high-throughput, managed production path for running Muse Spark Open across agentic coding and multimodal workloads. Start building: together.ai/models/muse-spar…

Muse Spark Open API: Pricing, Benchmarks & Docs | Together AI

Muse Spark Open, the coding-optimized open weights model from Meta, trained for agentic workflows with reliable tool calling, deep codebase understanding, multimodal perception, and a 1M-token...

together.ai
456
.@NVIDIA Nemotron 3.5 Super VL is now live on Together AI. An open-weight multimodal model built for visual reasoning and always-on agents that see, reason, and act across documents, screens, and video.
4
6
1
35
3,736
What makes NVIDIA Nemotron 3.5 Super VL different? → Text, image, document, and video understanding in one model → Reasoning and tool use with visual context preserved across steps → 120B hybrid MoE (12B active) → Multi-token prediction for faster generation → NVFP4 KV cache for more concurrent long-context sessions → Open-weights for customization and post-training
1
2
805
Run Nemotron 3.5 Super VL on Together AI Start building: together.ai/models/nemotron-…

NVIDIA Nemotron 3.5 Super VL API: Pricing, Benchmarks & Docs | Together AI

Open-weight multimodal MoE model with 120B parameters (12B active) for visual reasoning and long-running AI agents, combining document, image, and video understanding with tool use.

together.ai
1
509
Load more