---
title: "Free Inferencing: Free model APIs, verified"
description: "Find genuinely free LLM and model inference APIs, with current limits, model IDs, access requirements, and primary sources."
canonical_url: "https://freeinferencing.com/"
md_url: "https://freeinferencing.com/index.md"
last_updated: "2026-09-19"
---

# Free model APIs. Prototype freely.

> A source-backed catalog of hosted LLM, embedding, reranking, speech, image, safety, and research APIs with genuinely free access.

## Current snapshot

- **57** ongoing or recurring offers
- **7** community or conditional offers
- **49** trials or promotions
- **113** current offers in total
- **38** claims under review
- **47** records where no free hosted offer was verified

The snapshot is dated **2026-09-19**. Limits and model catalogs are volatile; re-check each provider's linked first-party sources before shipping.

## How to read the catalog

Free access is separated into recurring quotas, zero-price model routes, refreshing credits, research-only access, experimental access, community capacity, user-funded access, and finite trials. Prompt privacy is audited separately because zero price does not imply zero retention or zero training use.

Read the [methodology](/methodology.md) for the evidence rules and the [glossary](/glossary.md) for exact terminology.

## Evaluation checklist

- Separate recurring quotas and refreshing credits from signup trials, fixed promotions, and experimental capacity.
- Confirm the exact machine-readable model ID, current zero-price route, endpoint, authentication method, and supported request shape.
- Translate request, token, concurrency, daily, and monthly limits into the workload's actual peak traffic; the first ceiling reached is the practical allowance.
- Audit prompt retention, training, product improvement, human access, and downstream routing independently from price.
- Test one authenticated request with the intended account type, then inspect response metadata, usage counters, and balance movement before scaling.
- Exercise authentication, rate-limit, quota-exhaustion, and model-removal failures without allowing a silent switch to a billable route.
- Record the dated sources, chosen model, account conditions, privacy rationale, fallback, and owner responsible for re-verification.

The [watchlist](/watchlist.md) preserves plausible offers with a material unresolved term. The [no-free inference directory](/shame.md) preserves retired tiers, paid-only APIs, consumer chats, BYOK gateways, and self-hosted models so failed claims are not repeatedly presented as current provider-funded inference.

## Current providers

- [GroqCloud](/provider/groqcloud.md): Always-free quota. A recurring no-cost API allowance covering 13 cataloged models.
- [Hetzner Experiments Inference API](/provider/hetzner_experiments.md): Experimental access. Free while the provider's public experiment remains active.
- [Cloudflare Workers AI](/provider/cloudflare_workers_ai.md): Always-free quota. A recurring no-cost API allowance covering 8 cataloged models.
- [OpenRouter](/provider/openrouter.md): Free model routes. 25 model routes currently listed at zero price.
- [Jina Reader and anonymous utility APIs](/provider/jina_ai_reader.md): Always-free quota. A recurring no-cost API allowance covering 2 cataloged models.
- [AwanLLM](/provider/awanllm.md): Always-free quota. A recurring no-cost API allowance covering 6 cataloged models.
- [IBM watsonx.ai Runtime](/provider/ibm_watsonx_ai_runtime.md): Always-free quota. A recurring no-cost API allowance covering 4 cataloged models.
- [Maritaca AI Academic Credits](/provider/maritaca_academic_credits.md): Research / development. No-cost research and development inference across 10 models; not unrestricted production use.
- [Modal](/provider/modal.md): Monthly credits. $30 in credits refresh on a published schedule.
- [Nous Research / Nous Portal](/provider/nous_portal.md): Free model routes. 7 model routes currently listed at zero price.
- [Arli AI](/provider/arliai.md): Always-free quota. A recurring no-cost API allowance with provider-published access terms.
- [ModelScope API-Inference](/provider/modelscope_inference.md): Research / development. No-cost research and development inference across 1 model; not unrestricted production use.
- [Cohere](/provider/cohere.md): Research / development. No-cost research and development inference across 10 models; not unrestricted production use.
- [Albert API](/provider/albert_api_france.md): Research / development. No-cost research and development inference across 10 models; not unrestricted production use.
- [Lightning AI Model APIs / litAI](/provider/lightning_ai_model_apis.md): Monthly credits. $15 in credits refresh on a published schedule.
- [Sail Research](/provider/sail_research.md): Monthly credits. $5 in credits refresh on a published schedule.
- [Z.AI Model API](/provider/zai.md): Free model routes. 3 model routes currently listed at zero price.
- [Google AI Studio / Gemini Developer API](/provider/google_gemini_api.md): Always-free quota. A recurring no-cost API allowance covering 22 cataloged models.
- [Vercel AI Gateway](/provider/vercel_ai_gateway.md): Monthly credits. $5 in credits refresh on a published schedule.
- [Swiss AI Research Platform Model API](/provider/swiss_ai_research_platform.md): Always-free quota. A recurring no-cost API allowance covering 33 cataloged models.
- [Api.Airforce](/provider/api_airforce.md): Free model routes. 28 model routes currently listed at zero price.
- [Hugging Face Inference Providers](/provider/huggingface_inference_providers.md): Monthly credits. $0.1 in credits refresh on a published schedule.
- [ETH Zurich SPH LLM API](/provider/eth_sph_llm_api.md): Always-free quota. A recurring no-cost API allowance covering 5 cataloged models.
- [Requesty LLM Gateway](/provider/requesty.md): Free model routes. 12 model routes currently listed at zero price.
- [Mistral Studio / La Plateforme](/provider/mistral.md): Monthly credits. $10 in credits refresh on a published schedule.
- [Arnict](/provider/arnict.md): Monthly credits. $5 in credits refresh on a published schedule.
- [Qdrant Cloud Inference](/provider/qdrant_cloud_inference.md): Always-free quota. A recurring no-cost API allowance covering 6 cataloged models.
- [ElevenLabs API](/provider/elevenlabs_api.md): Always-free quota. A recurring no-cost API allowance covering 5 cataloged models.
- [GWDG KISSKI Chat AI and SAIA](/provider/gwdg_saia.md): Research / development. No-cost research and development inference across 16 models; not unrestricted production use.
- [NVIDIA Build / NVIDIA-hosted NIM](/provider/nvidia_build.md): Research / development. No-cost research and development inference across 82 models; not unrestricted production use.
- [Ollama Cloud](/provider/ollama_cloud.md): Monthly credits. credits refresh on a published schedule.
- [LLM7](/provider/llm7.md): Always-free quota. A recurring no-cost API allowance covering 2 cataloged models.
- [OVHcloud AI Endpoints](/provider/ovhcloud_ai_endpoints.md): Free model routes. 7 model routes currently listed at zero price.
- [BazaarLink](/provider/bazaarlink.md): Free model routes. 3 model routes currently listed at zero price.
- [Cartesia](/provider/cartesia.md): Always-free quota. A recurring no-cost API allowance covering 3 cataloged models.
- [OrcaRouter](/provider/orcarouter.md): Free model routes. 4 model routes currently listed at zero price.
- [Cerebras Inference](/provider/cerebras_inference.md): One-time trial. $5 for new accounts; it does not recur.
- [Lelapa AI Vulavula API](/provider/lelapa_vulavula.md): Always-free quota. A recurring no-cost API allowance with provider-published access terms.
- [NSF National Deep Inference Fabric](/provider/ndif.md): Research / development. No-cost research and development inference across 114 models; not unrestricted production use.
- [OpenTyphoon API](/provider/opentyphoon.md): Research / development. No-cost research and development inference across 4 models; not unrestricted production use.
- [Beam](/provider/beam_cloud.md): Monthly credits. $30 in credits refresh on a published schedule.
- [Goethe University AI-ToolLab LLM API](/provider/goethe_ai_toollab.md): Always-free quota. A recurring no-cost API allowance covering 17 cataloged models.
- [TU Wien Aqueduct AI Gateway](/provider/tu_wien_aqueduct.md): Always-free quota. A recurring no-cost API allowance covering 8 cataloged models.
- [Arizona State University Research Computing LLM API](/provider/asu_rc_llm_api.md): Always-free quota. A recurring no-cost API allowance with provider-published access terms.
- [Argonne Leadership Computing Facility Inference Endpoints](/provider/alcf_inference_endpoints.md): Research / development. No-cost research and development inference across 7 models; not unrestricted production use.
- [Kilo AI Gateway](/provider/kilo_ai_gateway.md): Free model routes. 22 model routes currently listed at zero price.
- [LLM.API](/provider/llmapi_ai.md): Free model routes. 1 model route currently listed at zero price.
- [DreamPrompting](/provider/dreamprompting.md): Always-free quota. A recurring no-cost API allowance covering 7 cataloged models.
- [Jetstream2 LLM Inference Service](/provider/jetstream2_llm.md): Research / development. No-cost research and development inference across 3 models; not unrestricted production use.
- [OCI Generative AI](/provider/oracle_oci_generative_ai.md): One-time trial. $300 for new accounts; it does not recur.
- [Scaleway Generative APIs](/provider/scaleway_generative_apis.md): One-time trial. 1M tokens for new accounts; it does not recur.
- [Aalto Local LLM Gateway](/provider/aalto_local_llm_gateway.md): Research / development. No-cost research and development inference across 1 model; not unrestricted production use.
- [AI Horde](/provider/ai_horde.md): Community capacity. Community-supplied inference across 26 models; capacity varies and throughput is not guaranteed.
- [Deepgram](/provider/deepgram.md): One-time trial. $200 for new accounts; it does not recur.
- [Fikra API](/provider/fikra_api.md): One-time trial. 300K tokens for new accounts; it does not recur.
- [Novita AI](/provider/novita_ai.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [Palabra.ai API](/provider/palabra_ai_api.md): One-time trial. $50 for new accounts; it does not recur.
- [Pollinations.ai](/provider/pollinations.md): Free model routes. 21 model routes currently listed at zero price.
- [Tencent Hunyuan](/provider/tencent_hunyuan.md): One-time trial. 1M tokens for new accounts; it does not recur.
- [BytePlus ModelArk](/provider/byteplus_modelark.md): One-time trial. 500K tokens for new accounts; it does not recur.
- [ch.at](/provider/ch_at.md): Community capacity. Community-supplied inference across 1 model; capacity varies and throughput is not guaranteed.
- [Jina Search Foundation API](/provider/jina_ai_search_foundation.md): One-time trial. 10M tokens for new accounts; it does not recur.
- [Speechmatics](/provider/speechmatics.md): One-time trial. $100 for new accounts; it does not recur.
- [Waterfall](/provider/waterfall.md): Free model routes. 27 model routes currently listed at zero price.
- [AI21 Studio](/provider/ai21_studio.md): One-time trial. $10 for new accounts; it does not recur.
- [Fireworks AI](/provider/fireworks_ai.md): One-time trial. $1 for new accounts; it does not recur.
- [Helmholtz Blablador](/provider/helmholtz_blablador.md): Research / development. No-cost research and development inference across 14 models; not unrestricted production use.
- [National Research Platform Managed LLMs](/provider/nsf_nrp_managed_llm.md): Research / development. No-cost research and development inference across 10 models; not unrestricted production use.
- [WestAI / KI:Inferenz.nrw LLM Hosting](/provider/westai_ki_inferenz_nrw.md): Research / development. No-cost research and development inference across 6 models; not unrestricted production use.
- [Hyperbolic](/provider/hyperbolic.md): One-time trial. $1 for new accounts; it does not recur.
- [AI.cc / AICC](/provider/ai_cc.md): One-time trial. $1 for new accounts; it does not recur.
- [Baidu Qianfan](/provider/baidu_qianfan.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [Google Cloud Vertex AI](/provider/google_vertex_ai.md): One-time trial. $300 for new accounts; it does not recur.
- [Puter.js AI](/provider/puter_js.md): User pays. Free to the developer; each end user signs in and consumes their own allowance.
- [Tencent HY Text Generation Global / Hunyuan Translation](/provider/tencent_hunyuan_global_translation.md): One-time trial. 100M tokens for new accounts; it does not recur.
- [WaveSpeedAI](/provider/wavespeedai.md): One-time trial. $1 for new accounts; it does not recur.
- [Amazon Bedrock](/provider/aws_bedrock.md): One-time trial. $100 for new accounts; it does not recur.
- [AMD Radeon Cloud Public Free Model APIs](/provider/amd_radeon_cloud_free_model_api.md): Research / development. No-cost research and development inference across 4 models; not unrestricted production use.
- [Replicate](/provider/replicate.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [FreeInference.org](/provider/freeinference_org.md): Always-free quota. A recurring no-cost API allowance covering 8 cataloged models.
- [Inception Platform](/provider/inception_platform.md): One-time trial. 100M tokens for new accounts; it does not recur.
- [Microsoft Foundry Models](/provider/azure_ai_foundry.md): One-time trial. $200 for new accounts; it does not recur.
- [Nebius Token Factory](/provider/nebius_token_factory.md): One-time trial. $1 for new accounts; it does not recur.
- [Virginia Tech ARC LLM Gateway](/provider/virginia_tech_arc_llm_api.md): Always-free quota. A recurring no-cost API allowance covering 14 cataloged models.
- [Poolside direct API](/provider/poolside_direct_api.md): Limited promotion. A limited-duration free promotion with no permanence guarantee.
- [SiliconFlow Global](/provider/siliconflow.md): One-time trial. $1 for new accounts; it does not recur.
- [Mancer AI](/provider/mancer_ai.md): Free model routes. 1 model route currently listed at zero price.
- [Lucidity Composite](/provider/lucidity_composite.md): Always-free quota. A recurring no-cost API allowance covering 28 cataloged models.
- [Alibaba Cloud Model Studio / Qwen API](/provider/qwen_cloud.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [MARA Inference Cloud](/provider/mara_inference_cloud.md): One-time trial. $5 for new accounts; it does not recur.
- [Sarvam AI](/provider/sarvam_ai.md): One-time trial. ₹100 for new accounts; it does not recur.
- [Unbiased Pareto](/provider/unbiased_pareto.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [Logfare](/provider/logfare.md): Always-free quota. A recurring no-cost API allowance covering 23 cataloged models.
- [AIRAWAT Indian Startup Credits](/provider/airawat_india_startup_credits.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [FastRouter](/provider/fastrouter.md): Free model routes. 20 model routes currently listed at zero price.
- [ILMU API Free Plan](/provider/ilmu_api.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [Viggle API](/provider/viggle_api.md): One-time trial. $0.01 for new accounts; it does not recur.
- [Voyage AI](/provider/voyage_ai.md): One-time trial. 200M tokens for new accounts; it does not recur.
- [AssemblyAI](/provider/assemblyai.md): One-time trial. $50 for new accounts; it does not recur.
- [Stability AI Developer Platform](/provider/stability_ai_api.md): One-time trial. $0.01 for new accounts; it does not recur.
- [Ramp Router](/provider/ramp_router.md): One-time trial. $26 for new accounts; it does not recur.
- [Neuralk Seldon API](/provider/neuralk_seldon_api.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [Entrim.ai](/provider/entrim_ai.md): One-time trial. $10 for new accounts; it does not recur.
- [Upstage](/provider/upstage.md): One-time trial. $10 for new accounts; it does not recur.
- [Clarifai](/provider/clarifai.md): One-time trial. $5 for new accounts; it does not recur.
- [OpenCode Zen](/provider/opencode_zen.md): Limited promotion. A limited-duration free promotion with no permanence guarantee.
- [Taiwan ADI Computing Power Program](/provider/taiwan_adi_ai_power.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [SEA-LION API](/provider/sea_lion_api.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [Public AI Inference Utility](/provider/public_ai.md): One-time trial. A finite allowance for new accounts; it does not recur.
- [Async Voice API](/provider/async_voice_api.md): One-time trial. $0.1 for new accounts; it does not recur.
- [Paxa Labs API](/provider/paxa_labs.md): One-time trial. $1000 for new accounts; it does not recur.
- [Arcee Open Models API Beta](/provider/arcee_open_models_api_beta.md): One-time trial. $5 for new accounts; it does not recur.
- [Inferon](/provider/inferon.md): One-time trial. A finite allowance for new accounts; it does not recur.

## Sitemap

See the full [semantic sitemap](/sitemap.md) for every page and markdown mirror.