Open models: a census of 126 listings

Ada internally hosts open-source LLMs for contextual understanding only; reply generation runs on the commercial providers and the customer selects neither.

Source retrieved 2026-09-02

Four local-runtime adapters plus a browser WebGPU adapter let open-weight models run on your own hardware with no hosted inference call at all.

Source retrieved 2026-09-09

Local runtimes llama.cpp, LM Studio, Ollama and vLLM are first-class provider keys, alongside Llama, Mistral, DeepSeek and Qwen.

Source retrieved 2026-09-02

Runs against open-weight models served locally through Ollama or LM Studio, and against any endpoint exposing an OpenAI-compatible API, so no inference need leave your machine.

Source retrieved 2026-09-09

Open-weight models are explicitly in scope and the routing layer picks between open and closed models on cost and latency; Llama 3 is named among the selectable third-party foundation models.

Source retrieved 2026-09-03

One open-weight model, GPT-OSS-120b, is offered as a hosted choice on the consumer plans and is switched off for Enterprise; you cannot supply your own weights.

Source retrieved 2026-09-10

Open-weight models run only as custom models on your own Groq, Cerebras or OpenAI keys; none appear in the pre-deployed list Live Hub hosts.

Source retrieved 2026-09-03

Open-weight GLM and Kimi models are offered, but only Augment-hosted; the model list is closed and has no self-supplied weights.

Source retrieved 2026-09-02

An Ollama client runs open-weight models locally, and the Azure AI Foundry client is documented against Phi-4; the Ollama section is labelled experimental.

Source retrieved 2026-09-02

A published case study describes shipping locally hosted open-weight models via Ollama, and the white-label page names Hugging Face as a generative AI platform they deliver on.

Source retrieved 2026-09-11

Open-weight models are reachable only through hosted providers Botpress maintains: Cerebras, Groq, Fireworks AI and Mistral AI. There are no local weights, no Ollama, Together or Hugging Face integration, and no self-hosted inference.

Source retrieved 2026-09-02

The LLM layer routes through LiteLLM to 100+ providers, but the vendor names only Anthropic, OpenAI and Google, and its own speech models are API-only.

Open-weight models are a first-class path: a curated set ships inside the ClinePass subscription, and Ollama, LM Studio or any OpenAI-compatible endpoint runs weights you host.

Source retrieved 2026-09-12

--oss runs Codex against a local Ollama or LM Studio provider.

Source retrieved 2026-09-02

The included basic model is GPT OSS 120B, an open-weight model, and it is free on every paid plan rather than a credit-metered extra.

Source retrieved 2026-09-05

Local and open-weight serving is documented end to end, with step-by-step Ollama setup plus Hugging Face, Together, Groq and local NVIDIA NIM examples via LiteLLM.

Source retrieved 2026-09-13

All inference is hosted: the catalogue includes GLM 5.2 from Z.ai and Kimi K3 from Moonshot, but there is no path to point Cursor at open weights you run.

Source retrieved 2026-09-02

Open-weight LLMs are listed as supported models: gpt-oss-20b through Groq, which requires you to supply the endpoint, and an NVIDIA Nemotron model that Deepgram manages. Deepgram's own speech and voice weights are not distributed.

Source retrieved 2026-09-02

Open-weight models including DeepSeek, Kimi and GLM are selectable in Devin CLI.

Source retrieved 2026-09-02

Open-weight models are hosted by Cognition and included from Pro; the model cost table carries DeepSeek, Moonshot (Kimi), Z.ai (GLM) and Nvidia entries. You cannot point it at your own weights.

Source retrieved 2026-09-03

Locally deployed models are supported on air-gapped Enterprise deployments; on Cloud you install a provider from the Marketplace and supply its credentials.

Source retrieved 2026-09-02

Open-weight families such as Mistral, DeepSeek, Kimi and GLM are offered, but Dust serves them itself through providers like Fireworks rather than your own weights.

Source retrieved 2026-09-09

Open-weight models run two ways: ElevenLabs-hosted Qwen in the native model list, or Llama and Mixtral through a custom LLM endpoint such as Together AI, Groq or SambaNova.

Source retrieved 2026-09-02

Droid Core is an open-weight model pool with its own rate limits, switchable free when standard usage runs out.

Source retrieved 2026-09-02

Ollama is a documented provider option, but only in the open-source column of the vendor's own comparison table. The matching Cloud cell reads 'Managed provider path', so a hosted customer cannot run an open-weight model.

Source retrieved 2026-09-03

The vendor confirms it serves open-weight models itself on US inference platforms, and its own Gen-1 slide model is post-trained from an open-weight base.

Source retrieved 2026-09-15

Two Moonshot AI open-weight models are in the hosted catalog, served by GitHub on Azure AI Foundry and Fireworks AI under zero data retention.

Source retrieved 2026-09-02

Open-weight models are a named tier in the rate card, served through Fireworks or Baseten.

Source retrieved 2026-09-02

Open-weight models run through LiteLLM against local hosts such as Ollama, with separate documented paths for vLLM and on-device LiteRT-LM serving.

Source retrieved 2026-09-15

Open-weight models run locally through Ollama, LM Studio, Docker Model Runner, Ramalama or goose's own local-models command, and remotely through gateways such as Novita, OVHcloud and Cerebras.

Source retrieved 2026-09-16

Local open-weight models run through Ollama, and the supported-provider list also includes huggingface, vllm, together and deepseek.

Source retrieved 2026-09-16

Open-weight models from DeepSeek, Z.ai, MiniMax and others run as agents, and Gumloop markets them as the cheaper default rather than a fallback.

Source retrieved 2026-09-09

Fine-tuned open-weight models handle the cheapest routing tier on IrisAgent's own infrastructure, and private deployments can run open models inside your boundary.

Source retrieved 2026-09-03

Ollama is supported for models running on your own hardware, and DeepSeek, Mistral, Z.AI and OpenRouter cover open-weight models hosted elsewhere; the provider list is explicitly extended by any OpenAI-compatible endpoint.

Source retrieved 2026-09-17

Junie Local ships a bundled local inference engine, and the CLI connects to Ollama, LM Studio or LiteLLM endpoints.

Source retrieved 2026-09-02

Open-weight hosted models plus fully local models through Ollama or LM Studio.

Source retrieved 2026-09-02

Open-weight models run locally through Ollama or hosted through Groq, Fireworks, Together, Baseten and OpenRouter.

Source retrieved 2026-09-02

Open-weight models run through local servers: Ollama, vLLM and LM Studio each have components that point at an instance you operate.

Source retrieved 2026-09-03

Open-weight models run locally through the ChatOllama integration, and self-hosted vLLM and Hugging Face endpoints are documented as chat-model integrations.

Source retrieved 2026-09-02

Any MCP client you write yourself can drive the product, so the driving model is yours to choose; the vendor never names an open-weight model and its own hosted generation is not selectable.

Source retrieved 2026-09-02

Ollama, LM Studio and llama.cpp endpoints are local-only integrations; DeepSeek, Mistral and Together are also listed.

Source retrieved 2026-09-02

Open-weight models are the recommended default in LiveKit Inference (Gemma 4 31B, GPT OSS 120B, DeepSeek), and the Ollama plugin runs models locally.

Source retrieved 2026-09-02

The open-source framework integrates HuggingFace and local models; the hosted platform's enumerated model families are OpenAI, Anthropic and Google only.

Source retrieved 2026-09-02

A locally hosted model can be configured, but the vendor never names an open-weight family anywhere I read; only OpenAI, Anthropic and Azure OpenAI are identified by name.

Source retrieved 2026-09-09

Open-weight models run against any OpenAI-compatible endpoint on your own hardware; the vendor names gpt-oss, Qwen3 and DeepSeek.

Source retrieved 2026-09-02

Ollama, Foundry Local and ONNX Runtime GenAI run models locally, with reduced capability: Ollama function tools depend on the chosen model and the provider table gives ONNX no tool support at all.

Source retrieved 2026-09-02

Open-weight families are in the hosted catalog and, separately, any OpenAI-API-compatible checkpoint you run yourself can be registered and selected inside a block.

Source retrieved 2026-09-02

Ollama chat and embedding nodes run local open-weight models; OpenRouter, Hugging Face Inference, DeepSeek and Mistral nodes cover hosted open-weight options.

Source retrieved 2026-09-02

Open-weight families are selectable through the OpenRouter and Bedrock entries in the catalogue, so you are not confined to closed frontier models.

Source retrieved 2026-09-09

Any OpenAI-compatible endpoint or the beta Any-LLM/LiteLLM adapters can be used, which covers self-hosted open-weight serving, but no open-weight model or local runtime is named in the docs.

Source retrieved 2026-09-02

Local and self-hosted inference servers are first-class providers, as is any OpenAI- or Anthropic-compatible endpoint.

Source retrieved 2026-09-02

Open-weight models run locally through Ollama, LM Studio or llama.cpp, and the Go subscription is built specifically around open-source models.

Source retrieved 2026-09-02

Open-weight models are recommended and benchmarked by the vendor, with documented setup for serving them locally through LM Studio, Ollama, vLLM and SGLang.

Source retrieved 2026-09-02

Open-weight Nemotron, DeepSeek, GLM and Kimi models are callable by id and hosted by Perplexity; the Router API is a further open-weight catalogue in private preview.

Source retrieved 2026-09-02

Open-weight families such as GLM, DeepSeek and Kimi are offered in the catalogue, but they run on Pickaxe's infrastructure and are billed in credits, so you are not running the weights yourself.

Source retrieved 2026-09-09

The model catalogue enumerates only Raven, OpenAI and Amazon Bedrock (Claude, Nova) with no open-weight option, but you can attach your own LLM endpoint.

Source retrieved 2026-09-02

Open-weight models run through the first-party Ollama, Hugging Face, Groq, Cerebras and OpenRouter integrations, including a local Ollama server.

Source retrieved 2026-09-02

Open-weight models reachable through the OpenRouter integration; the vendor also hosts LLaMA-family models in its own cloud

Source retrieved 2026-09-02

Open-source model providers are a real but second-class category for Agent: they exist as a provider policy and are off unless an admin turns them on.

Source retrieved 2026-09-02

The hosted model menu enumerates only OpenAI, Anthropic and Google models, so no open-weight model is offered, but the custom LLM WebSocket path exists precisely so you can host one yourself.

Source retrieved 2026-09-02

Rox offers to integrate a customer's own trained models on request; it names no open-weight model and exposes no way to run one from the product.

Source retrieved 2026-09-03

The Salesforce-managed model list names only Anthropic, OpenAI/Azure OpenAI, Google Vertex, Amazon and NVIDIA (beta), with no Llama or Mistral, so open-weight models arrive only through the customer's own BYOLLM connection.

Source retrieved 2026-09-03

Open-weight models are part of the mix Sierra runs alongside frontier and its own proprietary models; Sierra decides which serves each turn.

Source retrieved 2026-09-03

Open-weight models run through Groq/Cerebras/OpenRouter, or locally via Ollama or vLLM at no API cost.

Source retrieved 2026-09-03

The LLM is entirely yours, so an open-weight model works; the vendor publishes no list of supported models because it never runs one for you.

Source retrieved 2026-09-03

Open-weight models are reachable two ways: Llama-family models through a Cerebras key, and any model you host yourself through the Local provider connection.

Source retrieved 2026-09-09

Open-weight Qwen and Moonshot Kimi families run on Telnyx's own GPUs and are the recommended default for assistants, so prompt and response generation never leaves Telnyx infrastructure.

Source retrieved 2026-09-09

No open-weight family appears in the closed docs table of models supported for Agents, which lists only Google, OpenAI and Anthropic; open-source models are offered only for on-premises Automation Suite, plus a generic connector for any OpenAI-V1-compatible endpoint.

Source retrieved 2026-09-03

Open-weight models reach the pipeline through Groq, Together AI, Mistral, DeepSeek and OpenRouter, or a self-hosted server.

Source retrieved 2026-09-03

Kimi and DeepSeek sit in the hosted catalogue, but the vendor never describes any model as open-weight and publishes no way to supply or run your own.

Source retrieved 2026-09-03

Inference can be pointed at a local Ollama instance or at Fireworks/OpenRouter with your own key; the built-in Budget profile runs DeepSeek V4 Flash.

Source retrieved 2026-09-03

Open-weight models are hosted for you through Fireworks AI, and an Auto (Open-weights) router picks between them; Llama is absent from that enumerated table.

Source retrieved 2026-09-03

Open-weight models run through llama.cpp, Ollama, LM Studio or any local OpenAI-compatible server, feeding the Zed Agent and Inline Assistant.

Source retrieved 2026-09-03

Open-weight models are not sold in the managed catalogue; you reach them by declaring your own Ollama or vLLM endpoint in settings.json and paying for that inference yourself.

Source retrieved 2026-09-09

Open weights run locally through Ollama, an OpenAI-compatible endpoint, or a managed llama.cpp router with GGUF download, load and unload from /llama.

Source retrieved 2026-09-03