Together AI
Together AI serves Muse Glimmer as a managed API. There are no weights to download, no runtime to install, and no GPU to own. You send requests to https://api.together.xyz/v1/chat/completions with a model string and an API key.
This page requires an API key. That is the trade the self-hosted guides don't make; see Run inference for those options.
Everything under "What Together publishes" is taken from Together's Muse Glimmer model page(opens in new tab), read on 2026-08-11. Anything Together measured but doesn't publish there is marked as theirs.
The model string
Plaintextmeta-models/Muse-Glimmer-30B
That string is the whole integration. It is what Together lists as the endpoint on the model page, and it's what appears in all three of their published examples.
What Together publishes
| Endpoint | meta-models/Muse-Glimmer-30B |
| Provider | Meta |
| Type | Chat, Vision |
| Deployment | Serverless, Dedicated |
| Parameters | 30B |
| Context length | 128K+ |
| Input modalities | Text, Image |
| Output modalities | Text |
| Input price | $0.35 / 1M tokens ($0.04 / 1M cached) |
| Output price | $1.50 / 1M tokens |
| Released | August 10, 2026 |
| License | Apache 2.0 |
| Availability | 99.9% SLA, serverless and dedicated |
Tool calling is supported through the model's chat template, which makes this endpoint usable for the agent loops described in the tool-calling guide.
Get a key
bashexport TOGETHER_API_KEY=... # https://api.together.xyz/settings/api-keys
Check that it works and that the model is visible to your account:
bashcurl -s https://api.together.xyz/v1/models \-H "Authorization: Bearer $TOGETHER_API_KEY" | grep -o 'meta-models/Muse-Glimmer-30B'
Call it
Together publishes three clients. The curl form shows the wire format:
curlcurl -X POST "https://api.together.xyz/v1/chat/completions" \-H "Authorization: Bearer $TOGETHER_API_KEY" \-H "Content-Type: application/json" \-d '{"model": "meta-models/Muse-Glimmer-30B","messages": [{"role": "user", "content": "What are some fun things to do in New York?"}]}'
Python, using Together's official SDK (pip install together; it reads TOGETHER_API_KEY from the environment):
Pythonfrom together import Togetherclient = Together()response = client.chat.completions.create(model="meta-models/Muse-Glimmer-30B",messages=[{"role": "user", "content": "What are some fun things to do in New York?"}],)print(response.choices[0].message.content)
TypeScript, using together-ai:
JavaScriptimport Together from 'together-ai';const together = new Together();const completion = await together.chat.completions.create({model: 'meta-models/Muse-Glimmer-30B',messages: [{ role: 'user', content: 'What are some fun things to do in New York?' }],});console.log(completion.choices[0].message.content);
The path, bearer header, and {model, messages} body use the OpenAI chat-completions wire format. An HTTP client pointed at https://api.together.xyz/v1 can call it.
Prove tool calling works
A response with text only proves the endpoint is up. A tool-calling round trip verifies the model integration: expect tool_calls in the response, not prose.
curlcurl -s https://api.together.xyz/v1/chat/completions \-H "Authorization: Bearer $TOGETHER_API_KEY" -H "Content-Type: application/json" \-d '{"model": "meta-models/Muse-Glimmer-30B","messages": [{"role": "user", "content": "Weather in Paris, Tokyo and Cairo?"}],"tools": [{"type": "function", "function": {"name": "get_weather","description": "Get current weather for a city","parameters": {"type": "object", "properties": {"city": {"type": "string"}},"required": ["city"]}}}]}'
A working call returns three parallel tool_calls entries, one per city, in a single assistant turn, with reasoning available separately on the response object.
You can also drive the same prompt from Together's Muse Glimmer playground(opens in new tab).
Behavior worth knowing
Sampling. Defaults come from the checkpoint: temperature 0.95, top_p 1.0. Greedy decoding loops on this model, so don't set temperature to 0.
Weights. Nothing here downloads them. For the model card or to run the same model locally, use meta-models/Muse-Glimmer-30B(opens in new tab) on Hugging Face.
What Together measured
These metrics are not on the model page. Together reported them through Mourya Vangala Srinivasa on 2026-08-10. They describe Together's serving stack rather than anything you configure.
| Metric | Value |
|---|---|
| TTFT p50 | 307 ms |
| Output tokens/s per request | 105 |
Together measured a dedicated 2xH100-80GB tensor-parallel replica under a sustained 0.4 QPS long-context replay: mean input 58,290 tokens (p99 approximately 127K), mean output 244 tokens, and 1-2 requests in flight. That is long-prompt agentic traffic, not a short-prompt microbenchmark. TTFT at 1K-token inputs is substantially lower.
Together also reports serving the model quantized to FP8: MLP-only blockwise e4m3, 128x128, with dynamic activations. Attention, embeddings, and the vision tower stay BF16. Together chose this configuration after checking quality against a BF16 reference on the same hardware: GPQA-Diamond 83.7 +/- 0.9 over six runs, against 83.5 on the model card. Precision is not something you select on this endpoint. The detail helps when you compare a hosted result with a local BF16 result.
Troubleshooting
| Symptom | Cause | Fix |
|---|---|---|
401 on every request | TOGETHER_API_KEY unset, or the key is from a different account | Re-export the key from https://api.together.xyz/settings/api-keys and rerun the /v1/models check above |
| Output repeats or loops | temperature 0 or greedy decoding | Use the checkpoint defaults (temperature 0.95, top_p 1.0) |
| Multi-tool prompts produce one call per turn instead of parallel calls | <|eom|> passed as a stop token, or a client that flattens tools into the prompt | Send tools as a top-level request field and leave stop unset |
| Long requests end at 32,768 output tokens | Together caps generation per request | Split the task; the cap guards runaway reasoning loops |
503 Service unavailable minutes after provisioning a dedicated endpoint | Replica still warming; Together reports approximately 10-15 minutes for graph compilation and FP8 kernel warmup | Retry with backoff until the first 200. Serverless is not affected |
Support
Next steps
Tune your requests with the Muse Glimmer prompting guide, or compare Together AI with the self-hosted runtimes.