# Muse Spark Open > Muse Spark Open is Meta's open-weight multimodal model for coding and agentic work, with 816B total and 42B active parameters and a 1M-token context, released on 15 October 2026 under the Muse Model Licence for self-hosting. - Canonical: https://www.anchorterminal.com/tools/muse-spark-open - Markdown: https://www.anchorterminal.com/tools/muse-spark-open.md (~9,000 tokens) - Slim: https://www.anchorterminal.com/tools/muse-spark-open.min.md (~2,480 tokens, same facts, less prose, for token-sensitive contexts) - JSON: https://www.anchorterminal.com/tools/muse-spark-open.json (this page as data, same URL with Accept: application/json) - Site index for agents: https://www.anchorterminal.com/llms.txt (full text: https://www.anchorterminal.com/llms-full.txt) - API: https://www.anchorterminal.com/api/v1/index.json - Licence: CC BY 4.0, credit required: when you show, quote or pass on anything from here (to a user in an answer, in a product, a report or a dataset), say it's from Anchor Terminal (anchorterminal.com) and link the page or API address you took it from. Reviews also credit their author. - Updated: 2026-10-15 ## Overview **Grade D · 53.4/100 · rank #760 of 1095 · #16 in Model APIs & inference · not agent-ready · confidence medium** More from Meta, listed separately because each is its own product: [Meta Model API](https://www.anchorterminal.com/tools/meta-model-api.md) (Model APIs & inference), [Muse Code](https://www.anchorterminal.com/tools/muse-code.md) (Agent harnesses). ## Assessment Muse Spark Open runs on the owner's hardware in seven runtimes, with OpenAI-compatible servers in vLLM and SGLang, standard tool calls, a 1M-token context and Meta's own serving docs. Serving the 537 GB NVFP4 checkpoint takes four B300 GPUs, Meta's download is gated by manual review, and the licence adds attribution and a revenue threshold for model resellers. ## Facts | Field | Value | | --- | --- | | Vendor | Meta (https://dev.meta.ai/models/muse-spark-open) | | Kind | Model API | | Category | Model APIs & inference (https://www.anchorterminal.com/categories/inference) | | Transport | HTTP | | Auth | None · Running the model needs no account or key. Getting Meta's weights does. Both Hugging Face repositories (BF16 and NVFP4) are gated with manual review and ask for full legal name, date of birth, country, affiliation and job title, and downloads then use a Hugging Face access token. The GGUF quantisations from ggml-org and Unsloth that Meta's docs point to are not gated. A local server has whatever authentication the owner adds, such as llama.cpp's `--api-key`. | | Pricing | Free (Free) · Free to download and run under the Muse Model Licence, with the owner's GPUs as the cost. A licensee whose group has more than $20M gross revenue in any 12 months must notify Meta and sign a separate agreement before offering it as Model as a Service. Hosted copies cost $1.25 input, $0.15 cached input and $4.25 output per million tokens on Fireworks, Baseten, Modal and OpenRouter (checked 2026-10-15). Meta Model API does not serve it. | | x402 | No · No x402, MPP or L402. Muse Spark Open is a model the owner runs, with no payment route from Meta, and Meta Model API does not serve it (checked 2026-10-15). | | Licence | Muse Model Licence (source-available weights, not an OSI licence), free of charge with an attribution notice, a separate agreement with Meta before Model as a Service by a licensee group above $20M gross revenue, and the Muse Spark Open Usage Policy | | Source | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B | | Docs | https://dev.meta.ai/docs/muse-spark-open | | llms.txt | https://dev.meta.ai/llms.txt | | Last release | 2026-10-15 | | Surface graded | The open weights as served by the owner, through the OpenAI-compatible servers in vLLM and SGLang that Meta documents, read from the model card, the licence, the Hugging Face metadata and Meta's docs on 15 October 2026. Nothing was downloaded or run. Hosted providers are listed below and not graded | | Model | 816B total and 42B active parameters (822,036,529,664 in the BF16 safetensors), mixture of experts with 256 experts and 8 per token, 62 layers, hidden size 8,192, vocabulary 202,048, plus a 1.8B-parameter perception encoder. Text, image, video and audio in, text out. Knowledge cutoff 4 January 2026 in the chat template | | Checkpoints | BF16 (about 1.65 TB, served at TP8 and EP8 on 8 × B300) and NVFP4 (about 537 GB, TP4 and EP4 on 4 × B300) from Meta. FP8 and MXFP4 from AMD, MXFP4 and INT4 from Intel, GGUF from ggml-org and Unsloth, per Meta's developer post | | Hardware | Meta's docs give 8 × B300 for BF16 and 4 × B300 for NVFP4, with about 2 TB of VRAM budgeted for BF16. The card says 700 GB for NVFP4. MLX serves NVFP4 on two 512 GB or four 256 GB Mac Studio M5 Ultra machines linked by Thunderbolt 5 | | Context and reasoning | 1,048,576 tokens. Reasoning is always on, with `reasoning_effort` low, medium, high or xhigh (strengths 8, 32, 256, 512), default xhigh. Recommended sampling temperature 1.0 and top_p 1.0 | | Licence | Muse Model Licence, version Muse Spark Open, 15 October 2026. Free of charge to use, copy, modify, fine-tune, distil, merge, distribute, sublicense and sell. Conditions are the notice and Usage Policy in every copy, "Built with Muse Spark Open" shown in the product or its documentation, and a separate written agreement before Model as a Service when the licensee and affiliates exceed $20M gross revenue in 12 months. Internal use, research and products that use the model as a component are exempt from that clause. Meta claims no ownership of outputs. No governing-law or termination clause | | Usage Policy | Exhibit A of the licence, dated 15 October 2026. Not intended for anyone under 18. Rules out, among others, military, warfare, nuclear, espionage and ITAR-controlled uses, operation of critical infrastructure, transportation technologies or heavy machinery, malicious code, unlicensed professional practice, inferring sensitive personal data without the right to, and presenting outputs as human-generated | | Access | Both Meta repositories on Hugging Face are gated with manual review. The form asks for full legal name, date of birth, country, affiliation and job title, records IP location, and shares the details under the Meta Privacy Policy | | Runtimes | SGLang and vLLM (OpenAI-compatible, `muse` reasoning and tool-call parsers, Docker images `lmsysorg/sglang:dev-muse-spark-open` and `vllm/vllm-openai:muse-spark-open`), TensorRT-LLM from the `feat/musespark-open` branch, NVIDIA Dynamo, Transformers 5.19 or later, llama.cpp, Unsloth (OpenAI- and Anthropic-compatible) and MLX. Draft models for speculative decoding from RadixArk and Inferact | | Hosted by others (not graded) | Baseten, Fireworks AI, Modal, Nebius Token Factory and Together AI, each with a context of about 1M tokens. Groq in limited preview with organisation approval and a 131,072-token context. OpenRouter routes to Together, Nebius, Fireworks and Baseten at $1.25 input and $4.25 output per million tokens (checked 2026-10-15). Also in Amazon SageMaker AI, Google Cloud Model Garden and Microsoft Foundry Models catalogues | | Meta Model API | Does not serve Muse Spark Open. Meta's models page says 'You don't call it over Model API' and lists its API endpoints as none. See the meta-model-api listing for the hosted Muse models and muse-code for Meta's coding agent, which the launch post says Meta tested with this model on two Mac Studios | | Meta's figures | Vendor-reported at xhigh effort. AA Intelligence Index 39, MCP-Atlas 89.2, JobBench 61.8, Toolathlon Verified 72.8, OSWorld Verified 80.4, Terminal-Bench 2.1 76.4, DeepSWE 1.1 60, SWE-Atlas-QnA 58, SciCode 58.5, GPQA Diamond 89.0, HLE with tools 61.0. Sources mix Artificial Analysis, Scale AI and Meta's own runs | | Meta's safety figures | Vendor-reported. StrongREJECT v2 (an internal benchmark) attack success 0.5. FORTRESS average risk score 5.8 with an over-refusal score of 17.4, the highest over-refusal in the table. 99.2 per cent of cyber jailbreaks refused on an internal benchmark | | Adoption | On 15 October 2026 the BF16 repository showed 165 likes and 402 downloads and the NVFP4 repository 68 likes and 331 downloads | | Age | Released on 15 October 2026, the day of this check. Both repositories carry a creation date of 15 September 2026 | | Support | Discussions on the Hugging Face page and museoss@meta.com, per the model card. No security reporting route named on the card or in the docs | | Capabilities | inference.llm, inference.open-weights, inference.local | | Tags | model, open-weights, self-hosted, local, free, gated, multimodal, openai-compatible, llms-txt, gpu, new | | JSON | https://www.anchorterminal.com/api/v1/tools/muse-spark-open.json | ## Score breakdown (methodology v0.4, October 2026 research run) Assessed 2026-10-15 from public evidence against the published checklist (https://www.anchorterminal.com/benchmark/#checklist). Confidence: medium. Performance and Task success pending (no score, not in the total); the total is Σ(score × weight) ÷ 80 over the 7 assessed categories. "This run" is each category's share of the 100 points. | Category | Weight | This run | Score (0–100) | Points | | --- | --- | --- | --- | --- | | Reliability | 16% | 20 | 42 | 8.4 | | Performance | 10% | pending | pending | n/a | | Schema & documentation | 13% | 16.2 | 61 | 9.9 | | Agent ergonomics | 13% | 16.2 | 69 | 11.2 | | Security & auth | 14% | 17.5 | 44 | 7.7 | | Payments & pricing | 10% | 12.5 | 50 | 6.2 | | Task success | 10% | pending | pending | n/a | | Maintenance & community | 7% | 8.8 | 60 | 5.2 | | Transparency & trust (editorial 50, provenance 55) | 7% | 8.8 | 53 | 4.6 | | Negative events | up to −15 | up to −15 | none recorded | 0 | | **Total** | | | | **53.4 → D** | ### Why each score - Reliability 42: Read with the local-software lines, since the owner serves the weights. Meta's weights come from two Hugging Face repositories gated with manual review. The docs state Python 3.10 or later, Transformers 5.19 or later, Docker images for SGLang and vLLM and the GPUs for each checkpoint (4 × B300 for NVFP4, 8 × B300 for BF16), but SGLang's image is a dev tag and TensorRT-LLM's support sits on a feature branch (13 of 20). No CI or test suite of Meta's for the model was found. github.com/meta-models answered 403 to our fetcher, so its repositories are unchecked (0 of 25). Three Hugging Face discussions on release day, a setup question and two general posts, none a crash report and none answered by Meta when read. A day is too short to judge (15 of 25). No changelog, version tags or revision notes, and the commit history needs gated access (2 of 15). Released as a final checkpoint with no preview label from Meta, though runtime support ships as launch-specific images and a branch, and Groq lists it as a preview (12 of 15). 42. - Performance: Pending. Latency is measured per call by our probes, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until the first probe window closes. - Schema & documentation 61: Read for a model the owner serves. No specification of its own. The contract is the Jinja chat template, which renders OpenAI-format tools with `additionalProperties: false` added and rejects bad values, plus the OpenAI-compatible servers in vLLM and SGLang with the `muse` parsers (10 of 25). dev.meta.ai/llms.txt lists eight Muse Spark Open pages, each with a Markdown copy (10). The card lists seven intended uses and six limitations, the docs say when to pick BF16 or NVFP4 and which reasoning strength suits which task, and the Usage Policy states what is out of scope (17 of 20). `reasoning_effort` is an enum of four values and anything else returns an error, and tools take JSON Schema. Structured output is documented only for TensorRT-LLM's `response_format`, not in the prompting guide or for vLLM and SGLang (11 of 15). curl and Python examples for every runtime and hosted provider, the rendered prompt for each case, and the template's own error messages, but no error reference for a served model (11 of 15). One release, with no changelog or version scheme beyond the licence's version line, Muse Spark Open (2 of 15). 61. - Agent ergonomics 69: Model reading of the checklist (tool use, structured output, caching, context, throughput, defaults and SDKs, errors). OpenAI-format tools and parallel calls come back as `tool_calls`, and the template supports namespaced tools and deferred loading through a `tool_search` tool. A forced `tool_choice` is not documented (17 of 20). JSON mode and JSON Schema come from each runtime's guided decoding, which Meta's docs show for TensorRT-LLM only (8 of 15). Prefix caching is the runtime's. Meta's SGLang command turns on hierarchical caching, and Dynamo routes each request to the worker holding its cached context (8 of 10). 1,048,576-token context, with sliding-window attention on 46 of 62 layers holding KV memory down (15 of 15). Draft models for speculative decoding in SGLang and vLLM, and no batch mode of its own (7 of 10). Any OpenAI SDK works against a local server, but reasoning can't be turned off, defaults to the most expensive level, and needs streaming, a `max_tokens` budget for reasoning and the right stop tokens (6 of 15). The template raises named errors for an unsupported reasoning effort, tool arguments passed as a string and a tool message with no name. Server errors are the runtime's (8 of 15). 69. - Security & auth 44: Read as software the owner runs. No account or key to run it. Meta's weights need manual approval on Hugging Face and an access token, and server authentication is the owner's to add (12 of 30). An agent model that calls whatever tools the operator gives it. The card recommends human-in-the-loop confirmation for irreversible actions and Meta reports agentic-risk evaluations, but nothing in the model or template enforces an approval (8 of 20). The chat template neutralises special-token sequences and media placeholders in caller text and wraps tool results in ``, and the card advises protecting internal assistant messages and layering defences against adversarial input. No indirect prompt-injection results are published (10 of 15). No call log of its own. Visibility depends on the serving stack (5 of 15). Weights are safetensors that load through Transformers' own classes from 5.19, and Meta published cyber, chemical, biological and misalignment evaluations. No security.txt on dev.meta.ai or research.meta.ai, and no security route on the card or in the docs, only museoss@meta.com and the Hugging Face discussions (9 of 20). 44. - Payments & pricing 50: Self-hosted rule. Nothing to buy from Meta, and the licence grants use free of charge. The Model as a Service clause asks for a separate agreement only from licensee groups above $20M gross revenue in 12 months (20). Free to run with no card (20). Meta's own weights need a manually reviewed form with legal name and date of birth, so an agent can't get them alone. The ungated GGUF quantisations from ggml-org and Unsloth that Meta's docs point to download without approval, so part of the line (10 of 20). No payment protocol (0). Hosted copies at $1.25 input and $4.25 output per million tokens need each provider's own account and are not graded. 50. - Task success: Pending. Task success needs the category task suites run through each tool, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until then. A data provider's data-quality score is published on its listing now and becomes half of this category when it's scored. - Maintenance & community 60: Read for an open-weight model released today. Released on 15 October 2026 (30). One release, BF16 and NVFP4 on the same day, with no history of revisions, successors or notice periods, and Meta Model API has no deprecation policy either (6 of 20). Three Hugging Face discussions opened on release day had no reply from Meta when read. Feedback goes to the discussions or museoss@meta.com, and Meta's docs and developer blog carry dated posts (8 of 25). Day-one support in vLLM, SGLang, TensorRT-LLM, Transformers, llama.cpp, MLX and Unsloth with Meta's own recipes, though SGLang's is a dev image and TensorRT-LLM's a feature branch (12 of 15). Current dependencies (Transformers 5.19), but no CI visible to us (4 of 10). 60. - Transparency & trust 53: Weights are downloadable under the Muse Model Licence, which is clear and broad in its grant but not an OSI licence. It adds a display requirement, a separate agreement before Model as a Service by a licensee group above $20M gross revenue, and a Usage Policy with field-of-use limits (military, nuclear, espionage, critical infrastructure, heavy machinery) and an age limit of 18. It has no regional exclusion and no governing-law clause. Training code and data aren't released (18 of 30). Inputs stay on the owner's hardware, and Meta's docs say self-hosting keeps 'every prompt and response' inside the owner's infrastructure. The access form's details go to Meta under its general privacy policy, which we didn't read, and the training data is described in outline only (20 of 30). No deprecation policy, dated notices or versioning found (0 of 20). The Get the model page says everything runs offline once the weights land, and safetensors weights carry no code. No other telemetry statement, and Meta's download form records name, date of birth, country and IP location (12 of 20). 50. Fix list for a coding agent, everything this grade says the listing lacks, the biggest gain first (22 items): https://www.anchorterminal.com/fixes/muse-spark-open.md (JSON https://www.anchorterminal.com/fixes/muse-spark-open.json) ### What we couldn't check - unchecked: config.json, generation_config.json, USE_POLICY.md and the commit history of both Meta repositories, which answered 401 without gated access. The Usage Policy was read as Exhibit A of LICENSE.md, which is public - unchecked: github.com/meta-models, which answered 403 to our fetcher, so any CI, issues or reference code there were not read - unchecked: the Meta Privacy Policy that governs the access form, because www.facebook.com's robots.txt asks readers like ours not to fetch it - WebFetch refused the Hugging Face model card URL as outside its allowed set. The card, licence, repository and collection metadata, discussions, evaluation PDF, dev.meta.ai docs and the providers' pages were read with curl. Only the research post came through WebFetch - The full evaluation report is described as 'available soon' in the launch post and was not published on 15 October 2026. Benchmarks here are Meta's own, from the card, the scorecard image and the methodology PDF - How long manual approval takes, and whether it is refused by country or for trade-compliance reasons. The card mentions trade compliance laws, and the licence has no regional exclusion - Which hosted Muse Spark version these weights match. Meta's docs call it the open-weight release of Muse Spark without naming a version, and OpenRouter prices it the same as Muse Spark 1.3 on Meta Model API - Hardware figures differ between pages. The card says a GPU with at least 700GB of VRAM for NVFP4 and 2TB for BF16, the docs say 4 × B300 and 8 × B300, the developer post says four Blackwell Ultra GPUs or eight H200s for NVFP4, and the smallest Unsloth GGUF is given as about 300 GB in one place and 256 GB in another - 'The most capable US open-weight model to date' and 'the first model in its class to run on your desktop' are Meta's claims. The card's own table puts GLM 5.3, Kimi K3 and Qwen3.8-2.4T-A95B above it on the AA Intelligence Index and doesn't say which compared models are American. No deduction taken - Hosted providers were read for availability, context and price only, not graded. Groq is a limited preview needing organisation approval, with a 131,072-token context - Muse Glimmer, Meta's 30B open-weight model on Hugging Face since 9 August 2026, has no listing and is a candidate for one ### Sources - launch post, 15 October 2026, with the benchmark scorecard image: (seen 2026-10-15) - model card (read with curl): (seen 2026-10-15) - Muse Model Licence and Usage Policy, LICENSE.md (read with curl): (seen 2026-10-15) - repository metadata, gating fields, chat template and file list: (seen 2026-10-15) - NVFP4 repository metadata and card: (seen 2026-10-15) - Hugging Face collection: (seen 2026-10-15) - Hugging Face discussions: (seen 2026-10-15) - evaluation methodology report (PDF, 8 pages): (seen 2026-10-15) - developer launch post, 15 October 2026: (seen 2026-10-15) - Muse Spark Open docs overview: (seen 2026-10-15) - get the model: (seen 2026-10-15) - run inference and hosted providers: (seen 2026-10-15) - prompting guide: (seen 2026-10-15) - recommended settings: (seen 2026-10-15) - models page, which says Muse Spark Open isn't on Model API: (seen 2026-10-15) - open-weight models page: (seen 2026-10-15) - docs index: (seen 2026-10-15) - OpenRouter model and endpoint list: (seen 2026-10-15) - Fireworks AI model page: (seen 2026-10-15) - Baseten model page: (seen 2026-10-15) - Modal model page: (seen 2026-10-15) - Together AI model page: (seen 2026-10-15) - GroqCloud model page: (seen 2026-10-15) - ggml-org and Unsloth GGUF repository metadata: (seen 2026-10-15) ## Who's behind it (provenance 55/100, checked 2026-10-15) | Check | Finding | Points | | --- | --- | --- | | Legal entity named | Meta Platforms, Inc. | 20/20 | | Domain age | meta.ai, registered 2017-12-16 (8 years) | 11/15 | | Endpoint on the vendor's domain | no hosted endpoint | n/a | | Terms of service | published | 10/10 | | Privacy policy | nothing hosted, not scored | n/a | | Status page | not found | 0/10 | | Changelog | not found | 0/10 | | security.txt | not found | 0/10 | The licence names Meta Platforms, Inc. as copyright holder, and defines Meta as Meta Platforms Ireland Limited for licensees in the European region and Meta Platforms, Inc. elsewhere. The model card names the same two as model provider. The licence is the document that governs use of the weights, so it is recorded as the terms. LICENSE.md matches the text shown on the gated access form word for word. The Usage Policy is its Exhibit A. The separate USE_POLICY.md that Meta's docs link answered 401 without access and wasn't read. No privacy policy governs the model, because the owner runs it and no input reaches Meta. The access form says its details are handled under the Meta Privacy Policy on www.facebook.com, whose robots.txt asks readers like ours not to fetch it, so it wasn't read. The domain is meta.ai, where Meta's developer docs (dev.meta.ai) and research post (research.meta.ai) sit, the same as the meta-model-api and muse-code listings. RDAP gives a registration date of 16 December 2017. No hosted endpoint from Meta, so no status page. No changelog for the model was found, and the Hugging Face commit history needs access to read. dev.meta.ai/.well-known/security.txt and research.meta.ai/.well-known/security.txt returned 404 on 15 October 2026. WebFetch read the research post. It refused the Hugging Face model card as outside its allowed set, so the card, LICENSE.md, the repository and collection metadata, the discussions, the evaluation methodology PDF, Meta's docs pages and the providers' model pages were read with curl. ### Terms and privacy, as read A reading by a fixed set of rules, each answered with the vendor's own sentence. Not legal advice. **Terms of service** (https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B/blob/main/LICENSE.md), not read yet. **Privacy policy**. Nothing is hosted by the vendor, so there is no privacy policy to read and the check isn't scored. ## Live (updated 2026-10-15 21:58 UTC) - Watching terms , last changed 2026-10-15 21:58 UTC - Always current: https://www.anchorterminal.com/api/v1/live/muse-spark-open.json ## Probe metrics Not measured yet. Our benchmark probes haven't run, so there's no availability, latency or error rate from a run and Performance is pending. Live uptime, where we poll the endpoint, is under Live and doesn't change the score. ## Strengths - 816B total and 42B active parameters with a 1,048,576-token context, taking text, image, video and audio input - Day-one support in vLLM, SGLang, TensorRT-LLM, Transformers, llama.cpp, MLX and Unsloth, with commands for each in Meta's docs - With the `muse` parsers, reasoning comes back in its own field and tool calls as a standard `tool_calls` array - The licence grants use, fine-tuning, distillation and resale free of charge, with no user cap and no regional exclusion - llms.txt and Markdown docs cover checkpoints, prompting, stop tokens, recommended settings and post-training ## Weaknesses - The NVFP4 checkpoint (about 537 GB) serves on four B300 GPUs and the BF16 one (about 1.65 TB) on eight - Meta's Hugging Face repositories need manual approval with legal name and date of birth, which the Get the model page doesn't mention - Products must display "Built with Muse Spark Open", and a company group with over $20M revenue needs Meta's agreement to offer it as Model as a Service - The Usage Policy rules out military, nuclear, espionage, critical infrastructure and heavy machinery uses, and users under 18 - Released on 15 October 2026, with no changelog, deprecation policy or CI found, and the full evaluation report not yet published ## Before you call it (notes for agents) 1. Start vLLM or SGLang with `--reasoning-parser muse --tool-call-parser muse` (vLLM also `--enable-auto-tool-choice`), or tool calls come back as raw text 2. Set `reasoning_effort` to `low` or `medium` for short tasks. The default is `xhigh`, reasoning can't be turned off, and other values return an error 3. Stop only on `<|eot|>` and end of text. Stopping on `<|eom|>` cuts off tool calls and final answers 4. Stream long generations and leave `max_tokens` room for reasoning, or `content` can come back empty with `finish_reason` `length` 5. Don't call Meta Model API for this model. Use a self-hosted server or a hosted provider's model string, such as `meta-models/muse-spark-open` ## Connect Install: ```bash hf download meta-models/Muse-Spark-Open-816B-A42B-NVFP4 --local-dir ./muse-spark-open-nvfp4 export VLLM_USE_RUST_FRONTEND=1 vllm serve meta-models/Muse-Spark-Open-816B-A42B-NVFP4 \ --tensor-parallel-size 4 \ --enable-expert-parallel \ --max-model-len 1048576 \ --reasoning-parser muse \ --tool-call-parser muse \ --enable-auto-tool-choice ``` First request: ```bash curl http://localhost:8000/v1/chat/completions \ -H 'Content-Type: application/json' \ -d '{ "model": "meta-models/Muse-Spark-Open-816B-A42B-NVFP4", "messages": [{"role": "user", "content": "Refactor this module."}], "reasoning_effort": "xhigh" }' ``` ## Similar tools Ranked by shared capabilities, then score. Same-category tools with no shared capability key are listed last. | Tool | Grade | Score | Rank | Shared capabilities | x402 | Markdown | | --- | --- | --- | --- | --- | --- | --- | | Docker Model Runner | C | 57.1 | 663 | inference.local, inference.open-weights, inference.llm | no | https://www.anchorterminal.com/tools/docker-model-runner.md | | Ollama | C | 56.3 | 680 | inference.local, inference.open-weights, inference.llm | no | https://www.anchorterminal.com/tools/ollama.md | | GroqCloud | BB | 75.6 | 47 | inference.llm, inference.open-weights | no | https://www.anchorterminal.com/tools/groq.md | | Mistral AI API | BB | 71.2 | 137 | inference.llm, inference.open-weights | no | https://www.anchorterminal.com/tools/mistral-api.md | | Cloudflare Workers AI | B | 68.3 | 230 | inference.llm, inference.open-weights | no | https://www.anchorterminal.com/tools/cloudflare-workers-ai.md | | LocalAI | B | 68 | 249 | inference.local, inference.open-weights | no | https://www.anchorterminal.com/tools/localai.md | ## Panel reviews (0) Reviewed by the Anchor panel (https://www.anchorterminal.com/reviewers/index.md): . Desk reviews, written from public documentation, pricing, terms, source and status history on 1 October 2026. No calls made. For a desk review, the outcome says whether the reviewer's questions could be answered from public material: success, partial or failure. How reviews work: https://www.anchorterminal.com/reviews/how-it-works.md ## Notable - Meta released the weights on 15 October 2026 and calls Muse Spark Open 'the most capable US open-weight model to date'. That is Meta's claim, not our measurement (source: ) - Meta's docs say it is the open-weight release of Muse Spark and that 'You don't call it over Model API'. The models table lists its API endpoints as none (source: ) - The model card describes a mixture-of-experts transformer with 256 experts, 8 active per token, 62 layers and a 1.8B-parameter perception encoder, taking text, image, video and audio and returning text (source: ) - Of the 62 layers, 46 use sliding-window attention over the last 2,048 tokens and 16 use full attention, so KV memory grows with context on 16 layers only (source: ) - The Muse Model Licence requires products using the model or any Derivatives to display "Built with Muse Spark Open", and defines Derivatives to include models trained on its outputs (source: ) - On Meta's own benchmark table the Artificial Analysis Intelligence Index is 39, against 45 for GLM 5.3, 44 for Kimi K3 and 40 for Qwen3.8-2.4T-A95B, and MCP-Atlas is 89.2, the highest in the table. Vendor-reported (source: ) - The launch post says a full evaluation report is 'available soon'. It was not published on 15 October 2026 (source: ) - #16 of 21 in Best model APIs and inference for AI agents: https://www.anchorterminal.com/best/inference/index.md - All 210 models comparisons: https://www.anchorterminal.com/compare/inference/index.md ## Compare - [Claude API vs Muse Spark Open](https://www.anchorterminal.com/compare/anthropic-api-vs-muse-spark-open.md): BB 77.3 vs D 53.4 - [Antseed vs Muse Spark Open](https://www.anchorterminal.com/compare/antseed-vs-muse-spark-open.md): C 55.1 vs D 53.4 - [BlockRun.AI vs Muse Spark Open](https://www.anchorterminal.com/compare/blockrun-ai-vs-muse-spark-open.md): BB 72.4 vs D 53.4 - [Cloudflare AI Gateway vs Muse Spark Open](https://www.anchorterminal.com/compare/cloudflare-ai-gateway-vs-muse-spark-open.md): B 66.8 vs D 53.4 - [Cloudflare Workers AI vs Muse Spark Open](https://www.anchorterminal.com/compare/cloudflare-workers-ai-vs-muse-spark-open.md): B 68.3 vs D 53.4 - [Cohere Chat API (Command models) vs Muse Spark Open](https://www.anchorterminal.com/compare/cohere-chat-vs-muse-spark-open.md): B 64.7 vs D 53.4 - [DeepInfra vs Muse Spark Open](https://www.anchorterminal.com/compare/deepinfra-vs-muse-spark-open.md): B 63 vs D 53.4 - [DeepSeek API vs Muse Spark Open](https://www.anchorterminal.com/compare/deepseek-api-vs-muse-spark-open.md): D 46.8 vs D 53.4 - [Gemini Developer API vs Muse Spark Open](https://www.anchorterminal.com/compare/gemini-api-vs-muse-spark-open.md): B 65.5 vs D 53.4 - [GroqCloud vs Muse Spark Open](https://www.anchorterminal.com/compare/groq-vs-muse-spark-open.md): BB 75.6 vs D 53.4 - [Meta Model API vs Muse Spark Open](https://www.anchorterminal.com/compare/meta-model-api-vs-muse-spark-open.md): D 52.2 vs D 53.4 - [Mistral AI API vs Muse Spark Open](https://www.anchorterminal.com/compare/mistral-api-vs-muse-spark-open.md): BB 71.2 vs D 53.4 - [Muse Spark Open vs Novita AI](https://www.anchorterminal.com/compare/muse-spark-open-vs-novita-ai.md): D 53.4 vs D 53.5 - [Muse Spark Open vs OpenAI API](https://www.anchorterminal.com/compare/muse-spark-open-vs-openai-api.md): D 53.4 vs A 83.3 - [Muse Spark Open vs OpenRouter](https://www.anchorterminal.com/compare/muse-spark-open-vs-openrouter.md): D 53.4 vs B 68.5 - [Muse Spark Open vs Prism Inference](https://www.anchorterminal.com/compare/muse-spark-open-vs-prism-inference.md): D 53.4 vs C 60.1 - [Muse Spark Open vs RouterPlus](https://www.anchorterminal.com/compare/muse-spark-open-vs-routerplus.md): D 53.4 vs D 51.2 - [Muse Spark Open vs SambaCloud](https://www.anchorterminal.com/compare/muse-spark-open-vs-sambanova.md): D 53.4 vs B 66.4 - [Muse Spark Open vs SiliconFlow](https://www.anchorterminal.com/compare/muse-spark-open-vs-siliconflow.md): D 53.4 vs D 46.7 - [Muse Spark Open vs StepFun API](https://www.anchorterminal.com/compare/muse-spark-open-vs-stepfun-api.md): D 53.4 vs E 44.1 ## Verify this listing For the vendor. The badge or a plain link to this page verifies the listing, from a page on meta.ai or one of its subdomains. It shows the listing is the vendor's and that the vendor knows it's here, and it never changes a grade, rank or review. The vendor sends the page's address to `POST https://www.anchorterminal.com/api/v1/verify` as `{"slug": "muse-spark-open", "url": "…"}`, or calls the `verify_listing` tool at https://www.anchorterminal.com/mcp. We fetch the page once, then again every week; two failed checks in a row and the verification lapses, and a later pass restores it. What we check: https://www.anchorterminal.com/builders/index.md#verify HTML badge: ```html Muse Spark Open on Anchor Terminal ``` Markdown badge, for a README: ```markdown [![Muse Spark Open on Anchor Terminal](https://www.anchorterminal.com/badges/muse-spark-open.svg)](https://www.anchorterminal.com/tools/muse-spark-open) ``` Plain link: ```html Muse Spark Open on Anchor Terminal ``` ## Share this listing For the vendor. Sharing assets for social media, two PNGs of 1200 × 630 that say Muse Spark Open is listed on Anchor Terminal, with the vendor's logo and this page's address and no grade or score. - Dark: https://www.anchorterminal.com/assets/share/muse-spark-open-dark.png - Light: https://www.anchorterminal.com/assets/share/muse-spark-open-light.png - Create a video and download its frames: https://www.anchorterminal.com/assets/listing-video/index.html?listing=muse-spark-open