Free Inference You Can Build With

Free LLM inference you can build with: an API key, a CLI login, or inference bundled inside a coding tool or builder. The test is whether you can ship an artifact you keep — code, an app, a site, a design, an agent, a skill. Chat-only free tiers are excluded on purpose.

verified 2026-11-01 · about / contribute

Which free LLM API should I use?

Free LLM inference you can build with: 38 providers, 130 models, no card for most entries. Pick by how you get in:

Widest choice: Google AI Studio (Gemini API) — 16 free models in one place
Direct API key: OpenRouter — 15 free models, callable with a key
Coding tool: Kiro (AWS) — 4 free models, login access

Every provider is listed below, with a detail page for limits and how to get in.

Provider	Free tier	Free models	RPM	TPM	RPD	Day tokens	Context	Timeout	Notes
OpenCode Zen	Promo free (limited time)	Big Pickle, GLM-5.3-Flash Free, DeepSeek V4.1 Flash Free, MiMo-V2.6-Flash Free, Beam Free, Flint 40B Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Nemotron 3.5 Super VL Free, Jev 1.13 Free	—	—	—	—	—	Not published	Free models are limited-time promos; some may use data for training. No credit card. OpenAI-compatible base URL: https://opencode.ai/inference/openai/v1 (Jev 1.13 uses systemone endpoint). Free-tier chat models require the OpenCode client. Live CLI: opencode models \| grep -i free
OpenRouter	Rate-limited free	cohere/north-mini-code:free, dots-studio/dots-3-note-preview:free, google/gemma-4-26b-a4b-it:free, google/gemma-4-31b-it:free, liquid/lfm-2.5-2.6b:free, nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free, nvidia/nemotron-3-ultra-550b-a55b:free, nvidia/nemotron-3.5-content-safety:free, nvidia/nemotron-3.5-lightning:free, nvidia/nemotron-3.5-super-vl:free, recursive/flint-v1.0:free, thinkingmachines/inkling-small:free, thinkingmachines/inkling:free, xingchen-agi/xing4.0-29b-a4b:free, meta/muse-spark-1.3-contributor:free	—	—	—	—	—	Not published	OpenAI-compatible gateway (https://openrouter.ai/api/v1). 50 RPD on $0 accounts, 1,000 RPD with $10+ credits. No credit card required. Permanent $0 models have :free suffix.
Google AI Studio (Gemini API)	Rate-limited free	gemini-2.5-flash, gemini-2.5-flash-lite, gemini-3-flash-preview, gemini-3.1-flash-lite, gemini-3.1-flash-lite-preview, gemini-3.5-flash, gemini-3.5-flash-lite, gemini-3.6-flash, gemini-3.7-flash, gemini-3.8-flash, gemini-flash-latest, gemini-flash-lite-latest, gemini-omni-1.1-flash, gemini-omni-flash-preview, gemma-4-26b-a4b-it, gemma-4-31b-it	—	—	—	—	—	Not published	Free forever, no card, most generous major-provider tier. Live-verified against the v1beta API with an API key (2026-08-12) — model list and context windows are real; per-account quota varies by account age (see AI Studio rate-limit dashboard). Free-tier data may be used for training.
Groq	Rate-limited free	llama-3.3-70b-versatile, llama-3.1-8b-instant, meta-llama/llama-4-scout-17b-16e-instruct, openai/gpt-oss-120b, openai/gpt-oss-20b, qwen/qwen3.6-27b, qwen/qwen3.8-27b, meta-llama/llama-prompt-guard-2-22m, compound-beta	—	—	—	—	—	Not published	LPU hardware, fastest time-to-first-token. Limits pooled at organization level per model. No credit card required. Limits visible in x-ratelimit-* response headers.
Cerebras	Trial credits	gpt-oss-120b, qwen-3.8-27b	—	—	—	—	—	Not published	Wafer-scale hardware, 2,000+ tok/s. No permanent free tier — $5 trial credits, 30-day expiry, card required. Per-model limits on a rotating shortlist.
Cloudflare Workers AI	Rate-limited free	~80 free models (Llama 3.x/4, Qwen, Gemma, DeepSeek-R1 distills, FLUX, Whisper, BGE embeddings), @cf/meta/llama-3.3-70b-instruct-fp8-fast, @cf/deepseek-ai/deepseek-r1-distill-qwen-32b, @cf/qwen/qwen2.5-72b-instruct, @cf/meta/llama-3.1-8b-instruct	—	—	—	—	—	Not published	Neurons are normalized GPU-compute units, shared pool across all models. Resets 00:00 UTC. Hard stop when pool exhausted, errors not overage. No credit card.
Hugging Face	Rate-limited free	Llama 3.2 8B (Serverless), Qwen 2.5 7B (Serverless), Mistral 7B (Serverless), Inference Providers gateway (15+ partners)	—	—	—	—	—	Not published	Two products: Serverless API (free, rate-limited) and Inference Providers gateway. Cold starts 10-30s on unpopular models. Limits not published as fixed numbers.
Mistral La Plateforme	Rate-limited free (Experiment)	Mistral Large, Codestral, All API models (Experiment tier)	—	—	—	—	—	Not published	Phone (SMS) verification required, no card. Exact limits no longer published - see Admin Console Limits per workspace. Evaluation tier, not production.
SambaNova Cloud	Rate-limited free	DeepSeek-V3.1, DeepSeek-V3.2, Meta-Llama-3.3-70B-Instruct, gemma-4-31B-it, gpt-oss-120b	—	—	—	—	—	Not published	OpenAI-compatible base URL api.sambanova.ai/v1. Preview models can be pulled at any time. No credit card, no phone verification.
NVIDIA NIM (build.nvidia.com)	Trial credits	GLM-5, Kimi-2.5, NIM-packaged open models	—	—	—	—	—	Not published	Credit-based, not a rate-limited-free tier. Larger grants need corporate email, tie to ~90-day evaluation windows.
Z.AI (Zhipu)	Rate-limited free	GLM-5.1, GLM-4.5-Flash, GLM-4.7-Flash, GLM-4.6V-Flash (vision)	—	—	—	—	—	Not published	Free-tier limits revised twice in the past year - verify. Peak-hour throttling. Flash models are free regardless of tier. OpenAI-compatible.
Together AI	Trial credits	200+ open models (Llama, Qwen, DeepSeek)	—	—	—	—	—	Not published	No trial — paid entry, $5 minimum credit purchase, card required. Prepaid balance, no expiry. Startup programs can grant more.
DeepInfra	Trial credits	deepseek-ai/DeepSeek-V3.1, Qwen/Qwen3-32B, Qwen/Qwen2.5-72B-Instruct, zai-org/GLM-4.7	—	—	—	—	—	Not published	200 concurrent requests per model - high ceiling. Credit-based, no daily cap while credits last. Notable models below; 185+ total in catalog-wide trial credits.
GitHub Copilot (Free)	Rate-limited free	Auto-selected (GPT-5 mini, Claude Haiku/Sonnet, Gemini Flash line)	—	—	—	—	—	Not published	Free plan: 2,000 completions/mo plus an unpublished allowance of GitHub AI Credits; models auto-selected. Login with GitHub OAuth, no card.
Kiro (AWS)	Rate-limited free	Qwen3 Coder Next, DeepSeek 3.2, MiniMax M2.1, Claude Sonnet 4.5	—	—	—	—	—	Not published	Free plan: 50 credits/mo, resets monthly, unused credits do not roll over. Login via social or AWS Builder ID; no card for the free tier.
Trae (ByteDance)	Rate-limited free	Frontier models (bundled, auto-selected)	—	—	—	—	—	Not published	Free plan: 5,000 autocompletions/mo plus limited standard-queue usage and SOLO mode. Frontier models bundled in the vendor UI; no card.
Bolt.new	Rate-limited free	Frontier models (bundled, auto-selected)	—	—	—	—	—	Not published	Free plan: 1M tokens/mo with a 300K/day cap, resets monthly. Frontier models bundled; no card.
Lovable	Rate-limited free	Frontier models (bundled, Fable 5.1)	—	—	—	—	—	Not published	Free plan: 5 build credits/day (max 30/mo) plus 20 cloud and 4 AI credits/mo. Frontier models bundled (Fable 5.1); no card.
v0 (Vercel)	Rate-limited free	Frontier models (bundled, auto-selected)	—	—	—	—	—	Not published	Free plan: $5/mo credits plus 7 messages/day. Frontier models bundled; no card.
Kaggle Notebooks	Rate-limited free	Open-weight models you load (GPU runtime)	—	—	—	—	—	Not published	Free ~30 GPU-h/week (resets weekly), P100 or 2xT4, 12h sessions. Open-weight models you load yourself; phone verification may apply.
Hugging Face Spaces	Rate-limited free	Any Hugging Face model you load (ZeroGPU)	—	—	—	—	—	Not published	Static Spaces free for everyone; free personal accounts host up to 2 ZeroGPU Gradio Spaces. Any HF model you load; no card. Persistent or upgraded Spaces require a Pro subscription.
Cursor (Hobby)	Rate-limited free	Composer 2 (own), Auto (auto-selected, limited frontier)	—	—	—	—	—	Not published	Hobby plan: limited Agent requests and Tab completions; quota unpublished by the vendor. Own + limited frontier models bundled; no card.
Devin (Free)	Rate-limited free	SWE-1.7 / SWE-2 (own), Open models (auto-selected)	—	—	—	—	—	Not published	Free plan: light quota, unpublished. SWE-1.7/SWE-2 plus open models bundled; no card.
Windsurf	Rate-limited free	Frontier models (via Auto)	—	—	—	—	—	Not published	Free plan: light quota plus unlimited Tab/inline edits; numeric limits unpublished. Frontier models via Auto; no card.
Qoder	Rate-limited free	Frontier models + BYO key	—	—	—	—	—	Not published	Free plan: limited completions/NES and basic models, plus BYOK; quota unpublished. No card.
Antigravity CLI (Google)	Rate-limited free	Gemini 3.8/3.7/3.6 Flash, Gemini 3.1 Pro, Claude Sonnet/Opus 4.6, gpt-oss-120b	—	—	—	—	—	Not published	Individual plan: unlimited Tab and command requests with unpublished basic weekly rate limits. Google login; no card.
Google Colab	Rate-limited free	Gemini (AI features) + GPU for your code	—	—	—	—	—	Not published	Free GPU/TPU with 12h max sessions; Colab does not publish these limits. Google login; no card.
Figma (Starter)	Rate-limited free	Figma AI, Make, agent	—	—	—	—	—	Not published	Starter plan: 150 AI credits/day, up to 500/mo. Figma AI, Make, and agent bundled; no card.
Canva (Free)	Rate-limited free	Canva AI, Magic Media	—	—	—	—	—	Not published	Free plan: up to 20 Standard AI uses or 20 Premium per month (resets monthly); Magic Media text-to-image is 50 lifetime. No card.
Framer (Free)	Trial credits	Framer Agents	—	—	—	—	—	Not published	Free plan: 500 AI credits to try (one-off), free Framer domain, non-commercial. No card.
Replit (Starter)	Rate-limited free	Frontier models (bundled, auto-selected)	—	—	—	—	—	Not published	Starter: daily Agent allowance plus limited monthly cloud credits and 1 published app; numbers unpublished. Frontier models bundled; no card.
Kilo Gateway	Rate-limited free	cohere/north-mini-code:free, poolside/laguna-s-2.1:free, poolside/laguna-xs-2.1:free, liquid/lfm-2.5-2.6b:free, dots-studio/dots-3-note-preview:free, nvidia/nemotron-3.5-lightning:free, nvidia/nemotron-3.5-super-vl:free	—	—	—	—	—	Not published	Universal gateway, $0 forever for default free models. Anonymous keyless at 200 req/hour per IP; :free-tagged models. OpenAI-compatible. No credit card.
Nous Portal	Rate-limited free	poolside/laguna-s-2.1:free, poolside/laguna-xs-2.1:free, stepfun/step-3.7-flash:free, upstage/solar-mini4:free, reflection/beam-1.0-501b-a23b:free, ornith/ornith-1.5-122b:free	—	—	—	—	—	Not published	Nous Research Hermes models, $0 monthly plan. Free 50 RPM / 500k TPM. OpenAI-compatible. No card on free plan.
Cohere	Rate-limited free	Command A, Command R+, Command R7B	—	—	—	—	—	Not published	Evaluation key, 1,000 calls/month. Chat 20 req/min per model. Embed 2k inputs/min. No card for trial key. Non-production limits.
Requesty	Rate-limited free	nvidia/nemotron-3-super-120b-a12b, google/gemma-4-31b-it, poolside/laguna-xs.2	—	—	—	—	—	Not published	Gateway, $0 free models at 200 req/day. Routing, caching, EU residency. No credit card required.
OVHcloud AI Endpoints	Rate-limited free	10 free models (anonymous keyless)	—	—	—	—	—	Not published	EU-sovereign. Anonymous keyless 2 req/min per IP per model. Authed 400 req/min. No usage cap apart from rate. No card for anonymous.
SiliconFlow	Rate-limited free	deepseek-ai/DeepSeek-R1-Distill-Qwen-7B, deepseek-ai/DeepSeek-V3, Qwen/Qwen2.5-7B-Instruct, THUDM/glm-4-9b-chat	—	—	—	—	—	Not published	Large-scale inference gateway offering permanently free open-source model tiers plus 14M free signup tokens. OpenAI-compatible API. No credit card required.
DeepSeek Platform	Trial credits	deepseek-chat, deepseek-reasoner	—	—	—	—	—	Not published	5M free tokens granted on new account creation (valid 30 days). OpenAI-compatible API format with cache-miss/cache-hit pricing.