Eden AI vs Muse Spark Open

Eden AI 82

Router, read on the model lines. Bearer keys with per-key budgets, expiry and attribution, a key management API, and deny-by-default guardrails that limit models per member, role or key (25). The data retention page says request and response content isn't stored and data isn't used to train models by default (18). Content logging is opt-in, async results are deleted after 7 days, and an organisation can enforce Zero Data Retention and region rules; the EU endpoint refuses non-EU models with 451 (13). Cost and provider per response, and a monitoring dashboard by key (12). The security page says Eden AI is SOC 2 and ISO 27001 certified and links a trust centre; /.well-known/security.txt returns 404 and no disclosure policy was found (14). 25+18+13+12+14 = 82.

Muse Spark Open 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 <tool_output>, 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. dev.meta.ai/.well-known/security.txt answers 404 (re-read on 4 November), research.meta.ai per the 15 October check, and no security route is on the card or in the docs, only museoss@meta.com and the Hugging Face discussions (9 of 20). 12 + 8 + 10 + 5 + 9 = 44.

Eden AI 81

Router, scored on the API lines. OpenAPI 3.1 document for v3 (version 3.0.0, 46 paths) at api.edenai.run/v3/docs/openapi.json, plus a management API spec (25). llms.txt on the site and in the docs, llms-full.txt and Markdown pages (10). llms.txt states when to use Eden AI and when not to, and the docs separate chat models from expert models (17). Requests follow the OpenAI shape, and expert-model tools on the MCP server carry JSON Schema inputs (12). Examples throughout, but the spec documents only 200, 201, 202, 302 and 422, while llms.txt names 401 and 451 for region refusals (10). Versioned /v3 path, but no public changelog was found (7). 25+10+17+12+10+7 = 81.

Muse Spark Open 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 the Muse Spark Open overview and seven pages under it, plus three cookbook 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). 10 + 10 + 17 + 11 + 11 + 2 = 61.

Eden AI 30

No x402, MPP or L402 (0). Provider prices with no markup plus a 5.5% platform fee, published per model in the keyless catalogue (20). llms.txt says some models are billed $0 for tests, but whether a new account can call them without buying credits isn't stated (10). A person signs up in a browser and buys credits through Stripe (0). 0+20+10+0 = 30.

Muse Spark Open 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 Hugging Face form with legal name and date of birth, or, since the 4 November re-read of Get the model, a signed-in download from Meta's own page, 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 (the Unsloth repository is ungated on 4 November), 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. 20 + 20 + 10 + 0 = 50.

Eden AI 33

Router, read on the hosted lines. No status page is linked from the site or docs, and status.edenai.co didn't resolve for us on 6 November 2026 (0). No incident history to read (5). The rate-limits page gives 10 requests a second per account, shared across keys, while the plans page gives 7 a second, raisable to 15, so the figure is documented but inconsistent (10). Fallback retries the next model or provider on a 429 or error inside one request, but no client-facing 429 or Retry-After guidance was found (8). No SLA published; 'Dedicated support & SLA' is listed only for the custom Advanced plan, and the pricing page's '99.99%' uptime is a claim without a status page behind it (0). The v3 API is GA (10). 0+5+10+8+0+10 = 33.

Muse Spark Open 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). On 4 November the BF16 repository's discussions page lists 13 open threads from 15 October to 1 November, 10 discussions and 3 pull requests (one a draft). A script counted 29 replies across the seven threads with replies, six discussions and the draft pull request, none from the repository owner or a Meta member, and their text wasn't read, so crash reports weren't separated from other posts (15 of 25, held from 15 October and flagged). 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, and both repositories unchanged since 15 October, though runtime support ships as launch-specific images and a branch, and Groq lists it as a preview (12 of 15). 13 + 0 + 15 + 2 + 12 = 42.

Eden AI 75

Router, read on the model lines. Chat completions, Responses and an Anthropic Messages route in the OpenAI shape, with tool calling depending on the routed model (14 of 20) and structured output likewise (10 of 15). Caching is named in the terms as part of the service, without a per-model price table we read (8 of 15). Context depends on the model; the keyless /v3/models catalogue lists capabilities, prices and regions per model so an agent can choose (12 of 15). A Batches API (10 of 10). No SDK of its own; the docs use the OpenAI Python and TypeScript SDKs, and a hosted MCP server exposes about 40 expert-model tools plus upload_file, check_job and list_models (21 of 25). 14+10+8+12+10+21 = 75.

Muse Spark Open 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). 17 + 8 + 8 + 15 + 7 + 6 + 8 = 69.