Roboflow
Meta

Meta: Muse Glimmer 30B

Muse Glimmer 30B Overview

Muse Glimmer 30B is a dense vision language model from Meta built for long-horizon agentic work on local hardware. The architecture pairs a 52-layer causal text decoder with a roughly 1.8B parameter ViT-G/14 perception encoder for about 29.6 billion parameters in total, and it accepts interleaved text and image input so an agent can interpret screenshots, charts, and documents alongside conversation. The decoder uses grouped-query attention with 32 query heads and 2 key-value heads, a repeating pattern of three sliding-window local attention layers followed by one global layer, SwiGLU feed-forward blocks, and rotary position embeddings applied on the local layers, supporting a trained context of 131,072 tokens.

Meta describes the model as distilled from the larger Muse Spark and trained and evaluated around agentic behavior: end-to-end task completion, schema-accurate tool calling, multi-step reasoning across extended workflows, and recovery when a tool call returns an unexpected result. Reasoning effort is selectable across low, medium, high, and xhigh settings, and the model emits channel-scoped reasoning traces together with XML style tool calls rather than JSON, which requires parsers specific to this family. A companion block-diffusion drafter head predicts blocks of 16 tokens per forward pass for speculative decoding, with the main model verifying the proposals in parallel.

Muse Glimmer 30B Interactive Demo

Muse Glimmer 30B Details & Performance

Details

Resources

—

Vision Tasks

CaptioningChart Question AnsweringClassificationDocument Question AnsweringImage TaggingMulti-Label ClassificationOCRObject DetectionVision LanguageVisual Question Answering

Features

Foundation VisionLLMs with Vision CapabilitiesMultimodal Vision

Usage

Past 30 Days

Performance

Avg. Latency

Arena Rankings

Muse Glimmer 30B Vision Evals

Vision Evals is Roboflow's ground-truth benchmark: every model runs the same real-world samples across six vision tasks, and answers are scored against ground truth.

Evals updated August 12, 2026Pricing updated August 13, 2026

Overall score#14 of 28
70.8%
Avg cost / sample#6 of 28
$0.0013
Avg speed / sample#19 of 28
8.70s
Avg tokens / sample
2.3K

Strengths and weaknesses

Muse Glimmer 30B averages 70.8% across the six Vision Evals tasks, ranking #14 of 28 models overall.

Its weakest relative showing is Object Detection, ranking #20 of 28 at 41.0%.

At $0.0013 per sample it is the 6th cheapest of the 28 benchmarked models, and its average inference time of 8.7s per sample makes it the 19th fastest.

Performance profile

Field medianMuse Glimmer 30B

Field medians: Object Detection 55.2%, Counting 64.2%, Identification 84.4%, OCR 90.9%, Data Extraction 86.6%, Reasoning 58.0%.

Results by task

TaskScoreField (0 to 100)RankCost / sampleSpeed
Object Detection
41.0%
#20 of 28$0.002015.03s
Counting
66.2%
#12 of 28$0.00083.71s
Identification
81.3%
#16 of 28$0.00061.87s
OCR
92.1%
#10 of 28$0.00127.66s
Data Extraction
86.6%
#13 of 28$0.00072.17s
Reasoning
57.6%
#15 of 28$0.00106.55s

Price vs. performance

Score vs. cost

Overall benchmark score against estimated cost per sample. Upper-left is the sweet spot: high quality at low cost.

28 models on the current benchmark · scores and efficiency pooled across all six tasks at low effort · Muse Glimmer 30B highlighted

Muse Glimmer 30B scores from a single evaluation run · Methodology

View all Vision Evals →

Muse Glimmer 30B Pricing

Muse Glimmer 30B costs $0.350 per 1M input tokens and $1.50 per 1M output tokens.

Input$0.350 / 1M tokens
Output$1.50 / 1M tokens
Cached input$0.040 / 1M tokens

Pricing updated Aug 13, 2026

Alternatives to Muse Glimmer 30B

Other models worth comparing for similar use cases.

Qwen
Qwen3.5 122B A10B
Qwen3.5-122B-A10B is a high-capacity multimodal Mixture-of-Experts (MoE) model developed by Alibaba’s Qwen team as part of the Qwen3.5 model family. The architecture contains 122 billion total parameters while activating roughly 10 billion per token through sparse expert routing, allowing the model to balance large-scale reasoning ability with relatively efficient inference compared to dense models of similar size.The model is designed to process both text and visual inputs within a unified multimodal framework, enabling tasks that require reasoning across images, documents, charts, and natural language. This makes it suitable for applications such as document understanding, diagram interpretation, and complex visual question answering.Qwen3.5-122B-A10B supports a native context window of approximately 256,000 tokens, which can be extended further through techniques such as YaRN scaling to support very long-context workloads. Released under the Apache 2.0 license, it builds on earlier Qwen multimodal systems and provides developers with an open-weight model capable of handling demanding multimodal reasoning and analysis tasks.
Qwen
Qwen3 VL 235B A22B Instruct
Qwen3 VL 235B A22B Instruct is a flagship multimodal vision-language model developed by Qwen (Alibaba Cloud), designed for instruction-following tasks that combine advanced text generation with visual understanding. It serves as a high-end open-weight model for developers and researchers building multimodal AI systems that require strong reasoning, perception, and long-context capabilities.The model supports interleaved text and image inputs, very long context windows (up to roughly 256K tokens), and efficient inference through a mixture-of-experts architecture with about 22B active parameters out of 235B total. In today’s landscape, it competes with top-tier proprietary vision-language models while offering the advantages of open weights and flexible deployment. Typical applications include multimodal assistants, document and image analysis, visual reasoning, and large-context instruction-based workflows.
MoonshotAI
Kimi K3
Kimi K3 is a sparse Mixture-of-Experts large language model developed by Moonshot AI, with 2.8 trillion total parameters and a 1-million-token context window. The model activates 16 out of 896 experts per token using the Stable LatentMoE framework, and is built on two architectural innovations: Kimi Delta Attention (KDA), a hybrid linear attention mechanism that enables up to 6.3x faster decoding in long-context settings, and Attention Residuals (AttnRes), which selectively retrieves representations across model depth and delivers roughly 25% higher training efficiency. Together with refined training and data recipes, these structural advances yield approximately 2.5x better overall scaling efficiency compared to its predecessor Kimi K2. The model applies quantization-aware training from the supervised fine-tuning stage onward, using MXFP4 weights with MXFP8 activations for hardware compatibility. Thinking mode is always enabled at launch, with reasoning effort configurable via the reasoning_effort field.Kimi K3 supports native visual understanding alongside text, accepting image inputs for tasks that combine software engineering and visual reasoning. It targets long-horizon coding, knowledge work, and agentic workflows, and ships in two variants: K3 Max for general chat and agent tasks, and K3 Swarm Max for large-scale parallel processing across many coordinated sub-agents. The model is compatible with the OpenAI SDK via an OpenAI-compatible API. Full model weights are scheduled for release by July 27, 2026 under a Modified MIT license, following the open-weight pattern established by the Kimi K2 model family. A technical report with full architecture, training, and evaluation details is expected to accompany the weights release.
Meta
Muse Spark 1.1
Muse Spark 1.1 is a natively multimodal reasoning model from Meta Superintelligence Labs, released on July 9, 2026, as a significant upgrade to the original Muse Spark. The model accepts text, image, video, PDF, and audio as input and produces text output. It operates with a 1-million-token context window (1,048,576 tokens per the Meta Model API documentation) and is designed specifically for agentic tasks that require planning, tool use, computer use, and multi-agent orchestration. The model runs in a "Thinking" mode, where adjustable reasoning effort is applied before generating a response. It can function both as a main agent gathering context, forming plans, and delegating to parallel subagents and as a subagent that adheres to assigned tasks and escalates when needed. It is trained to decide autonomously when to write automation scripts versus interact directly with a user interface.Muse Spark 1.1 supports a range of multimodal capabilities including visual perception, image and video captioning, visual-to-code generation, and document analysis. The model was evaluated under Meta's Advanced AI Scaling Framework across frontier risk categories including chemical and biological threats, cybersecurity, and loss-of-control scenarios. Parameter count, architecture details, and training data composition are not publicly disclosed. The model is proprietary and closed-weight, accessible to consumers through the Meta AI app and to developers via the Meta Model API, which launched in public preview alongside this release.
Google
Gemini 3.1 Pro
Gemini 3.1 Pro is a proprietary multimodal model from Google’s Gemini 3 series, released in early 2026 and designed for advanced reasoning across large multimodal datasets. It accepts text, images, audio, video, and documents, supporting up to a 1-million-token input context with up to 64k output tokens. Compared with Gemini 3 Pro, it improves long-context synthesis and multi-step reasoning, enabling more reliable analysis of large documents, datasets, and software codebases.The model also advances visual understanding and grounding, allowing it to interpret UI screenshots, diagrams, and real-world scenes while referencing specific regions within images or video. These capabilities make Gemini 3.1 Pro well suited for multimodal workflows involving document processing, interface analysis, robotics research, and complex visual reasoning.
OpenAI
GPT-5.6 Sol
GPT-5.6 Sol is the flagship model in OpenAI's GPT-5.6 family, which also includes Terra (a balanced everyday-work tier) and Luna (a fast, cost-efficient tier). Sol is designed for demanding reasoning, long-horizon agentic workflows, software engineering, computer use, scientific research, and cybersecurity tasks. It introduces two new capability modes: a "max" reasoning effort setting that allocates additional compute time for difficult problems, and an "ultra" mode that coordinates multiple subagents in parallel to accelerate complex, multi-step work. The model supports native multimodal input, allowing it to process screenshots, diagrams, charts, documents, and photographs alongside text. A reported context window of approximately 1.5 million tokens enables processing of large codebases, lengthy research documents, and extended agentic sessions.GPT-5.6 Sol was announced on June 26, 2026, initially in a limited preview for trusted partners, and reached general availability on July 9, 2026. On the Agents' Last Exam benchmark, which evaluates long-running professional workflows across 55 fields, Sol scores 53.6. On Terminal-Bench 2.1, which tests command-line agentic coding workflows, Sol Ultra achieves 91.9%. The model also demonstrates gains in life sciences evaluations, including long-horizon genomics and quantitative biology analyses. OpenAI paired the release with its most extensive safety evaluation to date, combining human red teaming with large-scale automated testing, and classified Sol as High capability in both cybersecurity and biological risk under its Preparedness Framework, though it does not cross the Critical threshold in either category.

Deploy Muse Glimmer 30B with an API

Muse Glimmer 30B runs as a hosted REST endpoint through Roboflow Workflows. Pick a task, then hand the prompt to your coding agent or copy the code. Forking the workflow into a free Roboflow workspace replaces the your-workspace and YOUR_API_KEY placeholders with your own.

Connect your agent to Roboflow (once)

Add the Roboflow MCP server

claude mcp add --transport http roboflow https://mcp.roboflow.com/mcp

Run /mcp and authorize Roboflow in your browser when the OAuth flow opens.

Start a new Claude Code session so the MCP loads, then paste the prompt below (it works the same in any agent).

Fork this workflow to your Roboflow workspace to use it.

Integrate the Roboflow "Muse Glimmer 30B" workflow into my app.

- Endpoint: POST https://serverless.roboflow.com/<your-workspace>/workflows/playground-muse-glimmer-30b-c
- Auth: send my Roboflow API key as `api_key` in the request body, read from the ROBOFLOW_API_KEY env var (never hardcode).
- Body: { "api_key": ..., "inputs": { `image`: { type: "url" | "base64", value }, `model_api_key`: my provider key } }.

With the Roboflow MCP connected, call `workflows_get` on "playground-muse-glimmer-30b-c" to read the exact input schema and treat it as the source of truth. A live `workflows_run` for this workflow also needs my OpenRouter key (`model_api_key`) passed as a runtime parameter; if you don't have it yet, skip the test run — it will fail with a server error without the provider key, which is expected and not a problem with your code — and rely on the schema. Validate the real run via the REST call once the keys below are set. Without the MCP, use the contract above.

Before running the app, set up these keys so it does not error at runtime:
- `ROBOFLOW_API_KEY` (sent as `api_key`) from https://app.roboflow.com/settings/api
- `OPENROUTER_API_KEY` (sent as `model_api_key`) from https://openrouter.ai/keys — my OpenRouter key
Create a .gitignore'd .env with these variables, using placeholder values for any I haven't given you. Then pause and tell me directly, in your reply: the full path to the .env file, exactly which keys I need to paste in, and the link to get each one. Wait for me to confirm I've added them before you run anything. Do not run the app until I confirm.

Then add the integration to my codebase: match my project's language, framework, and conventions; read every key from environment variables (never hardcode); add basic error handling; and include a small runnable example. If you can't tell what language my project uses, ask me.
Installpip install inference-sdk

Fork this workflow to your Roboflow workspace to use it.

# 1. Import the library
from inference_sdk import InferenceHTTPClient

# 2. Connect to your workflow
client = InferenceHTTPClient(
  api_url="https://serverless.roboflow.com",
  api_key="YOUR_API_KEY"
)

# 3. Run your workflow on an image
result = client.run_workflow(
  workspace_name="your-workspace",
  workflow_id="playground-muse-glimmer-30b-c",
  images={
    "image": "YOUR_IMAGE.jpg"  # Path to your image file
  },
  parameters={
    "model_api_key": "YOUR_OPENROUTER_API_KEY"
  },
  use_cache=True  # cache workflow definition for 15 minutes
)

# 4. Get your results
print(result)

Fork this workflow to your Roboflow workspace to use it.

const response = await fetch('https://serverless.roboflow.com/your-workspace/workflows/playground-muse-glimmer-30b-c', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    api_key: 'YOUR_API_KEY',
    inputs: {
      "image": {"type": "url", "value": "IMAGE_URL"},
      "model_api_key": "YOUR_OPENROUTER_API_KEY"
    }
  })
});

const result = await response.json();
console.log(result);

Fork this workflow to your Roboflow workspace to use it.

curl --location 'https://serverless.roboflow.com/your-workspace/workflows/playground-muse-glimmer-30b-c' \
--header 'Content-Type: application/json' \
--data '{
  "api_key": "YOUR_API_KEY",
  "inputs": {
    "image": {"type": "url", "value": "IMAGE_URL"},
    "model_api_key": "YOUR_OPENROUTER_API_KEY"
  }
}'

Muse Glimmer 30B License

Apache-2.0 · Permissive license

Muse Glimmer 30B is released under Apache-2.0, a permissive license. The Muse Glimmer 30B license lets you run, fine-tune, and redistribute the model in commercial products with no obligation to open-source related code changes, so no separate commercial license is required.

Commercial use
Permitted. Because Apache-2.0 is permissive, Muse Glimmer 30B can ship inside paid products and internal systems with no commercial license and no revenue threshold.
Modification
Permitted. Fine-tuning, quantizing, and distilling are all allowed, and your code changes can stay closed. Files you change must be marked as changed.
Redistribution
Permitted with attribution. Ship the Apache-2.0 license text and any NOTICE file alongside the weights or derived code.

Apache-2.0 grants an express patent license that terminates if you bring a patent claim over the work, and it disclaims warranties. Validate Muse Glimmer 30B on your own data before you depend on it in production.

Read the full Apache 2.0 license ↗

Do I need a commercial license for Muse Glimmer 30B?

This is the straightforward case: a permissive license is the best technical solution and you are free to deploy Muse Glimmer 30B commercially without open-sourcing your own code.

Do not hesitate to reach out with questions for your commercial project — our team will help you start solving business problems on the first call. See Roboflow commercial licensing for the models included in each plan.

Talk to sales

This model is released under the Apache License 2.0, a permissive open-source license that allows commercial use, modification, distribution, and patent use.

Yes. Under the terms of the Apache 2.0 license, you can freely use this model for commercial purposes, including in proprietary products. You must retain the copyright notice and disclaimers when redistributing.

License information is provided as a guide and is not legal advice.

Frequently Asked Questions About Muse Glimmer 30B Vision

Yes. Muse Glimmer 30B accepts image input and handles OCR, data extraction, object counting, identification, visual reasoning, and object detection. On Roboflow's Vision Evals its strongest task is OCR at 92.1% (#10 of 28). You can test it on your own image in the demo above.

Yes. its transcriptions match the ground truth 92.1% on average (#10 of 28) on Vision Evals OCR. Pulling specific fields out of documents (data extraction) scores 86.6%.

Not its strength. On Vision Evals, Muse Glimmer 30B scores 41% mAP@50 on object detection (#20 of 28) and 66.2% exact-match accuracy on object counting. For production counting or precise localization, pairing it with a specialized detector like RF-DETR or your own trained model in a Roboflow Workflow is usually more reliable: detect the objects, then count the detections.

On our benchmark's task mix, Muse Glimmer 30B averages $0.0013 per sample at $0.35 per 1M input and $1.50 per 1M output tokens (#6 of 28 on cost), with an average speed of 8.7s per sample across the benchmark. Actual cost depends on your images and prompts.

On the overall Vision Evals ranking, Muse Glimmer 30B sits #14 of 28 at 70.8%, just behind GPT-5.6 Luna (71.5%) and just ahead of Gemini 3.5 Flash-Lite (69.6%). See the full side-by-side: Muse Glimmer 30B vs GPT-5.6 Luna.