{
  "listings": [
    {
      "slug": "koboldcpp",
      "name": "KoboldCpp",
      "company": "LostRuins (Concedo)",
      "grade": "B",
      "fit": "yes",
      "why": "MCP tool calling since 1.106 takes a Claude Desktop mcp.json with HTTP and STDIO servers, tool toggles and optional manual approval. An account had Devstral reformat headers through an MCP file-tools server.",
      "source": "https://github.com/LostRuins/koboldcpp/wiki",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/koboldcpp"
    },
    {
      "slug": "localai",
      "name": "LocalAI",
      "company": "Ettore Di Giacinto and the LocalAI team",
      "grade": "B",
      "fit": "yes",
      "why": "Functions map to grammars on llama.cpp with a parallel tool calls section and streamed reasoning separated from tool output. MCP connects HTTP and stdio servers with server-side execution limits.",
      "source": "https://localai.io/docs/features/openai-functions/",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/localai"
    },
    {
      "slug": "privategpt",
      "name": "PrivateGPT",
      "company": "PriBAI Technology Spain S.L.",
      "grade": "C",
      "fit": "yes",
      "why": "Built-in server tools, custom tools returned as Anthropic tool_use blocks, client tools (bash, text editor) and MCP connectors are documented with request shapes.",
      "source": "https://docs.privategpt.dev/api-guide/tools",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/privategpt"
    },
    {
      "slug": "text-generation-webui",
      "name": "TextGen",
      "company": "oobabooga",
      "grade": "E",
      "fit": "yes",
      "why": "Tool calls come back as finish_reason tool_calls with structured names and arguments; Qwen and Mistral call several in parallel. MCP servers over HTTP and stdio (mcp.json) are documented in the tutorial.",
      "source": "https://github.com/oobabooga/textgen/blob/main/docs/12%20-%20OpenAI%20API.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/text-generation-webui"
    },
    {
      "slug": "vllm",
      "name": "vLLM",
      "company": "The Linux Foundation (vLLM is a PyTorch Foundation project)",
      "grade": "D",
      "fit": "yes",
      "why": "Tool parsers per model family, named and required tool choice guaranteed parsable, parallel_tool_calls true by default, streaming parsers, and a Responses example with MCP tools.",
      "source": "https://docs.vllm.ai/en/stable/features/tool_calling/",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/vllm"
    },
    {
      "slug": "anythingllm",
      "name": "AnythingLLM",
      "company": "Mintplex Labs, Inc.",
      "grade": "E",
      "fit": "caveat",
      "why": "Agents take MCP servers over StdIO, SSE and Streamable HTTP on Docker and Desktop. An open issue shows the schema helper dropping a parameter named definitions or $defs while leaving it required, so the model guesses its shape.",
      "source": "https://docs.anythingllm.com/mcp-compatibility/overview",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/anythingllm"
    },
    {
      "slug": "berth",
      "name": "Berth",
      "company": "Trevin Tindall (T-Crypt on GitHub)",
      "grade": "B",
      "fit": "caveat",
      "why": "Tool calls pass through to the engine untouched; Berth adds queuing and cold starts of up to minutes, with 503 and Retry-After while an engine loads.",
      "source": "https://github.com/T-Crypt/berth/blob/main/docs/gateway.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/berth"
    },
    {
      "slug": "foundry-local",
      "name": "Foundry Local",
      "company": "Microsoft Corporation",
      "grade": "B",
      "fit": "caveat",
      "why": "Tool support differs by hardware variant: on a Snapdragon X2 NPU the model answered in plain text with no tool call and ToolChoice Required failed everywhere, with no fix at reading.",
      "source": "https://learn.microsoft.com/en-us/azure/foundry-local/reference/reference-rest",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/foundry-local"
    },
    {
      "slug": "hyperqwen",
      "name": "HyperQwen",
      "company": "syv.ai (Copenhagen, CVR DK44671247)",
      "grade": "C",
      "fit": "caveat",
      "why": "Tool calls go through vLLM's parser with a strict mode; the README notes a tool schema's maxLength drops decode to about 7 tokens a second without the bundled xgrammar fix.",
      "source": "https://github.com/syv-ai/HyperQwen/blob/main/README.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/hyperqwen"
    },
    {
      "slug": "jan",
      "name": "Jan",
      "company": "Menlo Research Pte Ltd",
      "grade": "C",
      "fit": "caveat",
      "why": "Stdio MCP servers with per-call approval and saved permissions; local models need tool calling enabled per model. An open issue on 0.8.8 shows a truncated tool call saved without arguments breaking the whole thread.",
      "source": "https://github.com/janhq/jan/blob/main/docs/src/pages/docs/desktop/mcp.mdx",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/jan"
    },
    {
      "slug": "lemonade",
      "name": "Lemonade",
      "company": "Advanced Micro Devices, Inc.",
      "grade": "B",
      "fit": "caveat",
      "why": "Chat completions return tool_calls and document streamed reassembly; tool_choice objects are treated as auto on llama.cpp models. The MCP endpoint exposes Lemonade as a server, not a client. See the 33k crash issue.",
      "source": "https://github.com/lemonade-sdk/lemonade/blob/main/docs/api/openai.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/lemonade"
    },
    {
      "slug": "llama-cpp",
      "name": "llama.cpp",
      "company": "ggml.ai, part of Hugging Face since 2026",
      "grade": "D",
      "fit": "caveat",
      "why": "Function calling has native and generic parsers, parallel_tool_calls off by default, and stdio MCP servers. One issue shows a tool call returned inside reasoning_content on Vulkan until a build fix.",
      "source": "https://github.com/ggml-org/llama.cpp/blob/master/docs/function-calling.md",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/llama-cpp"
    },
    {
      "slug": "lm-studio",
      "name": "LM Studio",
      "company": "Element Labs, Inc.",
      "grade": "C",
      "fit": "caveat",
      "why": "Tool use docs distinguish native and default support and warn that small models produce calls LM Studio cannot parse into tool_calls. MCP via API takes ephemeral servers or mcp.json with allowed_tools.",
      "source": "https://lmstudio.ai/docs/developer/openai-compat/tools",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/lm-studio"
    },
    {
      "slug": "mlx-lm",
      "name": "MLX LM",
      "company": "Apple Inc.",
      "grade": "C",
      "fit": "caveat",
      "why": "server.py builds tool_calls from the model's tool-calling template when the tokenizer supports it, but the documented request fields do not include tools, and SERVER.md calls the server unfit for production.",
      "source": "https://github.com/ml-explore/mlx-lm/blob/main/mlx_lm/server.py",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/mlx-lm"
    },
    {
      "slug": "ollama",
      "name": "Ollama",
      "company": "Ollama Inc.",
      "grade": "D",
      "fit": "caveat",
      "why": "Tool calling returns structured tool_calls, documents parallel calls and streaming reassembly. An issue shows qwen2.5-coder:7b writing the call as text and qwen3 calling a missing tool; the maintainer advised a better tool-use model.",
      "source": "https://docs.ollama.com/capabilities/tool-calling",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/ollama"
    },
    {
      "slug": "open-webui",
      "name": "Open WebUI",
      "company": "Open WebUI Inc.",
      "grade": "D",
      "fit": "caveat",
      "why": "Native MCP since 0.6.31 is Streamable HTTP only with OAuth 2.1, admin-added and scoped by group; stdio needs mcpo. Accounts use web search, code calling and mcpo tools with local 27B models.",
      "source": "https://docs.openwebui.com/features/extensibility/mcp",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/open-webui"
    },
    {
      "slug": "ghost-core",
      "name": "Core",
      "company": "ZMJ, Inc., doing business as Ghost, 325 9th Street, San Francisco, CA 94103, United States",
      "grade": "F",
      "fit": "unchecked",
      "why": "No API or tool documentation exists on ghost.ai; the setup guide covers pairing and Wi-Fi only.",
      "confidence": "low",
      "url": "https://www.anchorterminal.com/tools/ghost-core"
    },
    {
      "slug": "underdog",
      "name": "Underdog",
      "company": "Conway Research",
      "grade": "F",
      "fit": "unchecked",
      "why": "The model card says nothing about tool calling and underdog.ai refuses Claude-named crawlers in robots.txt, so the app docs were not read.",
      "confidence": "low",
      "url": "https://www.anchorterminal.com/tools/underdog"
    },
    {
      "slug": "bitrun",
      "name": "BitRun",
      "company": "Abhishek Khanra (IshekKhal on GitHub)",
      "grade": "E",
      "fit": "no",
      "why": "No tools parameter documented; research checkpoints the README says to check before relying on.",
      "source": "https://github.com/IshekKhal/bitrun/blob/main/README.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/bitrun"
    },
    {
      "slug": "docker-model-runner",
      "name": "Docker Model Runner",
      "company": "Docker, Inc.",
      "grade": "C",
      "fit": "no",
      "why": "The maintainer states one tool call per turn is an upstream grammar limit and two calls return a 500, so the two-calls-in-one-turn criterion cannot be met.",
      "source": "https://github.com/docker/model-runner/issues/1063",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/docker-model-runner"
    },
    {
      "slug": "gpt4all",
      "name": "GPT4All",
      "company": "Nomic, Inc.",
      "grade": "F",
      "fit": "no",
      "why": "No tools parameter on the API. An account reports JSON errors with remote models and the project asks whether development has stopped.",
      "source": "https://docs.gpt4all.io/gpt4all_api_server/home.html",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/gpt4all"
    },
    {
      "slug": "khoj",
      "name": "Khoj",
      "company": "Khoj Inc.",
      "grade": "E",
      "fit": "no",
      "why": "The functions list has built-in online search, code execution and image tools inside Khoj; no MCP or custom tool connection appears in the docs navigation.",
      "source": "https://docs.khoj.dev/features/chat",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/khoj"
    },
    {
      "slug": "localghost",
      "name": "LocalGhost",
      "company": "LocalGhost.ai Ltd (company number 17213100, registered in England and Wales, incorporated 12 May 2026)",
      "grade": "F",
      "fit": "no",
      "why": "No agent surface.",
      "source": "https://github.com/LocalGhostDao/localghost/blob/main/README.md",
      "confidence": "medium",
      "url": "https://www.anchorterminal.com/tools/localghost"
    },
    {
      "slug": "screenpipe",
      "name": "screenpipe",
      "company": "Negentropy Labs, Inc. (dba Screenpipe)",
      "grade": "C",
      "fit": "no",
      "why": "An MCP server, not a client or runner.",
      "source": "https://docs.screenpipe.com/mcp-server",
      "confidence": "high",
      "url": "https://www.anchorterminal.com/tools/screenpipe"
    }
  ],
  "marks": "yes: the docs show it does the job as the success test says. caveat: it does the job with a limit an agent should know. no: it doesn't do this job. unchecked: we couldn't read it.",
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "attributionRequired": true,
    "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-11-16",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.8",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run",
    "terms": "https://www.anchorterminal.com/terms/#data"
  },
  "useCase": {
    "accounts": 11,
    "after": [
      "agent-memory.store-and-confirm",
      "search.fetch-url"
    ],
    "area": "models",
    "asks": [
      "let my local model call a calculator tool",
      "connect an MCP server to a local model",
      "build a private assistant that runs Python in a sandbox",
      "give the model file tools so it can fix formatting in my headers"
    ],
    "before": [
      "decision-models.pick-tool-or-skill"
    ],
    "blockers": [
      "server returns 500 when the model makes two tool calls in one turn",
      "a truncated tool call is saved without arguments and every later message in the thread fails",
      "tool calling works on one hardware backend (CUDA) and not another (NPU)",
      "weak tool-use models such as 7B coder variants write the call as JSON text and the client never sees it"
    ],
    "capabilities": [
      "inference.local",
      "inference.tool-calling",
      "agent.mcp-client"
    ],
    "category": "local-ai",
    "categoryName": "Local AI",
    "checked": "2026-11-12",
    "companies": {
      "fit": 5,
      "fitWithLimit": 11,
      "tagged": 101
    },
    "competition": "crowded",
    "fits": [
      "koboldcpp",
      "localai",
      "privategpt",
      "text-generation-webui",
      "vllm"
    ],
    "fitsWithLimit": [
      "anythingllm",
      "berth",
      "foundry-local",
      "hyperqwen",
      "jan",
      "lemonade",
      "llama-cpp",
      "lm-studio",
      "mlx-lm",
      "ollama",
      "open-webui"
    ],
    "id": "local-ai.agent-tool-loop",
    "listings": {
      "caveat": 11,
      "no": 6,
      "notRead": 0,
      "tagged": 115,
      "unchecked": 2,
      "yes": 5
    },
    "name": "Run a non-coding agent with tools or MCP servers against a local model",
    "status": "draft",
    "success": "The runner returns tool calls in the structured tool_calls field and handles two or more calls in one turn. Measured: across ten turns every tool call comes back in that field with arguments that validate against the schema, and the conversation is never left in a state the client cannot continue.",
    "url": "https://www.anchorterminal.com/api/v1/use-cases/local-ai.agent-tool-loop.json",
    "weight": 15,
    "weightBasis": "11 of 50 accounts, most of them GitHub issues where the failure is the use case: tool calls parsed wrongly, dropped, or breaking the thread"
  }
}
