LLM APIs with permanent free tiers for text inference.

_{All endpoints are OpenAI SDK-compatible unless noted. Each link points to the provider's API key page.}

---

All of those free LLM APIs are available at manifest.build - make reliable agents.

Contents

• Provider APIs
• Inference providers
• Glossary

Provider APIs

APIs run by the companies that train or fine-tune the models themselves.

Aion Labs 🇮🇱

Permanent free tier, no credit card required. 15 RPM, 20K tokens/day. Specialized for roleplay and storytelling.

Base URL: https://api.aionlabs.ai/v1

Model Name	Context	Max Output	Modality	Rate Limit
aion-labs/aion-2.0	128K	32K	Text (reasoning)	15 RPM, 20K TPD
aion-labs/aion-rp-llama-3.1-8b	32K	32K	Text	15 RPM, 20K TPD
aion-labs/aion-3.0	128K	32K	Text (reasoning)	15 RPM, 20K TPD
aion-labs/aion-3.0-mini	128K	32K	Text (reasoning)	15 RPM, 20K TPD

Cohere 🇨🇦

Free "Trial" API key, no credit card. 1,000 API calls/month. Non-commercial use only.

Base URL: https://api.cohere.com/v2

Model Name	Context	Max Output	Modality	Rate Limit
Command A+ (218B)	128K	64K	Text + Image	20 RPM
Command A (111B)	256K	8K	Text	20 RPM
Command R+	128K	4K	Text	20 RPM
Command R	128K	4K	Text	20 RPM
Command R7B	128K	4K	Text	20 RPM
Command A Reasoning	256K	32K	Text (reasoning)	20 RPM
Command A Translate	8K	8K	Text	20 RPM
Command A Vision	128K	8K	Text + Image	20 RPM
Command R7B Arabic	128K	~4K	Text	20 RPM
Aya Expanse 32B	128K	4K	Text	20 RPM
Aya Vision 32B	16K	4K	Text + Image	20 RPM

Google Gemini 🇺🇸

Free tier, no credit card. Free-tier prompts may be used by Google to improve products. ¹

Base URL: https://generativelanguage.googleapis.com/v1beta

Model Name	Context	Max Output	Modality	Rate Limit
Gemini 3.7 Flash	1M	65K	Text + Image + Audio + Video	—
Gemini 3.6 Flash	1M	65K	Text + Image + Audio + Video	15 RPM, 1,500 RPD
Gemini 3.5 Flash	1M	65K	Text + Image + Audio + Video	15 RPM, 1,500 RPD
Gemini 3.5 Flash-Lite	1M	65K	Text + Image + Audio + Video	30 RPM, 1,500 RPD
Gemini 3.1 Flash-Lite	1M	65K	Text + Image + Audio + Video	30 RPM, 1,500 RPD
Gemini 2.5 Flash	1M	65K	Text + Image + Audio + Video	15 RPM, 1,500 RPD
Gemini 2.5 Flash-Lite	1M	65K	Text + Image + Audio + Video	30 RPM, 1,500 RPD
Gemini 2.5 Pro	1M	65K	Text + Image + Audio + Video	5 RPM, 50 RPD
Gemma 4 31B	256K	32K	Text	—
Gemma 4 26B A4B	256K	32K	Text	—

Mistral AI 🇫🇷

Free mode, enabled by default, no credit card required. $10/month in API credits, and free-mode prompts may be used to train Mistral models unless you opt out. ²

Base URL: https://api.mistral.ai/v1

Model Name	Context	Max Output	Modality	Rate Limit
Mistral Medium 3.5 (128B)	256K	—	Text + Image + Code	~1 RPS, 500K TPM
Mistral Small 4	256K	—	Text + Image + Code	~1 RPS, 500K TPM
Mistral Large 3	256K	—	Multimodal	~1 RPS, 500K TPM
Ministral 3 8B	256K	—	Text + Vision	~1 RPS, 500K TPM
Codestral	128K	—	Code	~1 RPS, 500K TPM
Ministral 3 3B	256K	—	Text + Vision	~1 RPS, 500K TPM
Ministral 3 14B	256K	—	Text + Vision	~1 RPS, 500K TPM

Z AI (Zhipu AI) 🇨🇳

Permanent free models, no credit card required. ³

Base URL: https://open.bigmodel.cn/api/paas/v4

Model Name	Context	Max Output	Modality	Rate Limit
GLM-4.7-Flash	200K	128K	Text (reasoning)	1 concurrent request
GLM-4.5-Flash (retirement announced)	128K	96K	Text (reasoning)	1 concurrent request
GLM-4.6V-Flash	128K	32K	Multimodal	1 concurrent request

Inference providers

Third-party platforms that host open-weight models from various sources.

Cloudflare Workers AI 🇺🇸

10,000 Neurons/day free, no credit card required. 75+ models available on the free tier. ⁴

Base URL: https://api.cloudflare.com/client/v4/accounts/{account_id}/ai/run

Model Name	Context	Max Output	Modality	Rate Limit
@cf/meta/llama-3.3-70b-instruct-fp8-fast	24K	Shared w/ context	Text	10K neurons/day (shared)
@cf/meta/llama-4-scout-17b-16e-instruct	131K	Shared w/ context	Multimodal	10K neurons/day (shared)
@cf/openai/gpt-oss-120b	128K	Shared w/ context	Text	10K neurons/day (shared)
@cf/google/gemma-4-26b-a4b-it	256K	Shared w/ context	Text + Vision	10K neurons/day (shared)
@cf/zai-org/glm-4.7-flash	131K	Shared w/ context	Text	10K neurons/day (shared)
@cf/mistralai/mistral-small-3.1-24b-instruct	128K	Shared w/ context	Text	10K neurons/day (shared)
@cf/deepseek-ai/deepseek-r1-distill-qwen-32b	80K	Shared w/ context	Text (reasoning)	10K neurons/day (shared)
+ 72 more models	Varies	Varies	Text, Image, Audio, Embeddings	10K neurons/day (shared)

Groq 🇺🇸

Free tier, no credit card. Ultra-fast LPU inference. ⁵

Base URL: https://api.groq.com/openai/v1

Model Name	Context	Max Output	Modality	Rate Limit
openai/gpt-oss-120b	131K	65K	Text	30 RPM, 1,000 RPD
openai/gpt-oss-20b	131K	65K	Text	30 RPM, 1,000 RPD
groq/compound	131K	8K	Text	30 RPM, 250 RPD
groq/compound-mini	131K	8K	Text	30 RPM, 250 RPD
qwen/qwen3.6-27b	131K	16K	Text	30 RPM, 1,000 RPD

Hugging Face 🇺🇸

$0.10/month in Inference Provider credits for free users (subject to change). Routes to Fireworks, Together, Hyperbolic, Nebius, Novita, DeepInfra and others. Thousands of models.

Base URL: https://router.huggingface.co/v1

Model Name	Context	Max Output	Modality	Rate Limit
Meta-Llama-3.1-8B-Instruct	128K	~4K	Text	Credit-metered
gemma-3-4b-it	131K	~4K	Text	Credit-metered
phi-4	16K	~4K	Text	Credit-metered
Qwen2.5-Coder-7B-Instruct	131K	~4K	Text	Credit-metered
Qwen2.5-7B-Instruct	131K	~4K	Text	Credit-metered
+ thousands of community models	Varies	Varies	Text, Image, Audio, Embeddings	Credit-metered

Kilo Code 🇺🇸

Free models with no credit card and no API key required. kilo-auto/free auto-router dynamically routes to models in the free pool. ⁶

Base URL: https://api.kilo.ai/api/gateway

Model Name	Context	Max Output	Modality	Rate Limit
nvidia/nemotron-3-ultra-550b-a55b:free	1M	65K	Text	200 req/hr
stepfun/step-3.7-flash:free	262K	262K	Text + Vision	200 req/hr
nvidia/nemotron-3-super-120b-a12b:free	262K	262K	Text	200 req/hr
nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free	256K	65K	Multimodal	200 req/hr
poolside/laguna-s-2.1:free	262K	32K	Text (code)	200 req/hr
poolside/laguna-xs-2.1:free	262K	32K	Text (code)	200 req/hr
cohere/north-mini-code:free	256K	64K	Text (code)	200 req/hr
openrouter/free	Varies	Varies	Text	200 req/hr
tencent/hy3:free	262K	128K	Text	200 req/hr
nvidia/nemotron-3.5-lightning:free	1M	65K	Text	200 req/hr
liquid/lfm-2.5-2.6b:free	64K	8K	Text	200 req/hr

LLM7.io 🇬🇧

API gateway with a free tier. Anonymous access needs no key and reaches the turbo models; a free token from token.llm7.io raises the rate and token limits but reaches the same models. ⁷

Base URL: https://api.llm7.io/v1

Model Name	Context	Max Output	Modality	Rate Limit
gpt-oss:20b	128K	—	Text	10 RPM, 60 req/hr (anonymous)
mistral-Nemo-Instruct-2407	128K	—	Text	10 RPM, 60 req/hr (anonymous)
minimax-m2.7	180K	—	Text (reasoning)	10 RPM, 60 req/hr (anonymous)

ModelScope 🇨🇳

Free API-Inference for registered users. Requires Alibaba Cloud account binding + real-name verification. ⁸

Base URL: https://api-inference.modelscope.cn/v1

Model Name	Context	Max Output	Modality	Rate Limit
Qwen/Qwen3.5-35B-A3B	256K	—	Text	2,000 RPD total; <=500 RPD/model (dynamic)
Qwen/Qwen3.5-27B	256K	—	Text	2,000 RPD total; <=500 RPD/model (dynamic)
+ API-Inference-enabled models	Varies	Varies	LLM, MLLM	Dynamic quotas + dynamic concurrency

NVIDIA NIM 🇺🇸

Free with NVIDIA Developer Program membership. 100+ models. Rate-limited per model.

Base URL: https://integrate.api.nvidia.com/v1

Model Name	Context	Max Output	Modality	Rate Limit
nvidia/nemotron-3-super-120b-a12b	1M	262K	Text	40 RPM, 10,000 RPD
nvidia/nemotron-3-nano-30b-a3b	262K	32K	Text	40 RPM, 10,000 RPD
nvidia/llama-3.1-nemotron-ultra-253b-v1	128K	4K	Text	40 RPM, 10,000 RPD
meta/llama-3.3-70b-instruct	128K	4K	Text	40 RPM, 10,000 RPD
mistralai/mistral-nemotron	128K	8K	Text	40 RPM, 10,000 RPD
google/gemma-4-31b-it	262K	8K	Text	40 RPM, 10,000 RPD
mistralai/mistral-large-2-instruct	128K	4K	Text	40 RPM, 10,000 RPD
minimaxai/minimax-m3	1M	~64K	Text	40 RPM, 10,000 RPD
nvidia/nemotron-3-ultra-550b-a55b	1M	262K	Text	40 RPM, 10,000 RPD
openai/gpt-oss-120b	131K	131K	Text	40 RPM, 10,000 RPD
openai/gpt-oss-20b	131K	131K	Text	40 RPM, 10,000 RPD
+ 92 more models	Varies	Varies	Text, Image, Video, Speech, Embeddings	40 RPM, 10,000 RPD

Ollama Cloud 🇺🇸

Free tier with usage limits. 16 cloud model families from the Ollama library. OpenAI SDK-compatible via https://ollama.com/v1. ⁹

Base URL: https://ollama.com/api

Model Name	Context	Max Output	Modality	Rate Limit
deepseek-v4-pro	1M	Model-dependent	Text	Session/weekly limits (unpublished)
deepseek-v4-flash	1M	Model-dependent	Text	Session/weekly limits (unpublished)
minimax-m3	512K	Model-dependent	Text	Session/weekly limits (unpublished)
kimi-k3	1M	Model-dependent	Text	Session/weekly limits (unpublished)
gpt-oss:120b	128K	Model-dependent	Text	Session/weekly limits (unpublished)
gpt-oss:20b	131K	Model-dependent	Text	Session/weekly limits (unpublished)
nemotron-3-ultra	262K	Model-dependent	Text	Session/weekly limits (unpublished)
mistral-large-3:675b	256K	Model-dependent	Text	Session/weekly limits (unpublished)
qwen3.5:397b	256K	Model-dependent	Text	Session/weekly limits (unpublished)
+ 7 more cloud models	Varies	Varies	Text	Session/weekly limits (unpublished)

OpenRouter 🇺🇸

17 free models (marked with :free suffix). OpenAI SDK-compatible. ¹⁰

Base URL: https://openrouter.ai/api/v1

Model Name	Context	Max Output	Modality	Rate Limit
nvidia/nemotron-3-super-120b-a12b:free	262K	262K	Text	20 RPM, 50 RPD
openai/gpt-oss-20b:free	131K	32K	Text	20 RPM, 50 RPD
cohere/north-mini-code:free	256K	64K	Text (code)	20 RPM, 50 RPD
google/gemma-4-26b-a4b-it:free	262K	32K	Text + Image	20 RPM, 50 RPD
google/gemma-4-31b-it:free	262K	32K	Text + Image	20 RPM, 50 RPD
inclusionai/ling-3.0-flash:free	262K	32K	Text	20 RPM, 50 RPD
nvidia/nemotron-3-nano-30b-a3b:free	256K	—	Text	20 RPM, 50 RPD
nvidia/nemotron-nano-9b-v2:free	128K	—	Text	20 RPM, 50 RPD
nvidia/nemotron-nano-12b-v2-vl:free	128K	128K	Text + Image	20 RPM, 50 RPD
poolside/laguna-s-2.1:free	262K	32K	Text (code)	20 RPM, 50 RPD
poolside/laguna-xs-2.1:free	262K	32K	Text (code)	20 RPM, 50 RPD
+ 6 more free models	Varies	Varies	Text / Image	20 RPM, 50 RPD

OVHcloud AI Endpoints 🇫🇷

Free anonymous tier (no API key, no signup): 2 RPM per IP per model. 20+ open-weight models hosted in EU. OpenAI SDK-compatible. ¹¹

Base URL: https://oai.endpoints.kepler.ai.cloud.ovh.net/v1

Model Name	Context	Max Output	Modality	Rate Limit
Qwen3.5-397B-A17B	131K	~32K	Text	2 RPM (anonymous)
gpt-oss-120b	128K	~32K	Text	2 RPM (anonymous)
gpt-oss-20b	128K	~8K	Text	2 RPM (anonymous)
Meta-Llama-3_3-70B-Instruct	131K	~4K	Text	2 RPM (anonymous)
Qwen3.6-27B	131K	~32K	Text	2 RPM (anonymous)
Qwen3.5-9B	131K	~8K	Text	2 RPM (anonymous)
Qwen3-32B	131K	~32K	Text	2 RPM (anonymous)
Qwen3-Coder-30B-A3B-Instruct	262K	~32K	Text (code)	2 RPM (anonymous)
Qwen2.5-VL-72B-Instruct	128K	~8K	Text + Vision	2 RPM (anonymous)
Mistral-Small-3.2-24B-Instruct	128K	~4K	Text	2 RPM (anonymous)
Mistral-Nemo-Instruct-2407	128K	~4K	Text	2 RPM (anonymous)
Mistral-7B-Instruct-v0.3	32K	~4K	Text	2 RPM (anonymous)

SiliconFlow 🇨🇳

Permanently free models, no credit card required. Identity verification required. 100+ models in the catalog, most of them paid. ¹²

Base URL: https://api.siliconflow.cn/v1

Model Name	Context	Max Output	Modality	Rate Limit
Qwen/Qwen3-8B	128K	—	Text	1,000 RPM, 50,000 TPM

Glossary

Abbreviation	Meaning
RPM	Requests per minute
RPD	Requests per day
TPM	Tokens per minute
TPD	Tokens per day
RPS	Requests per second

Contributing

Know a free tier that's missing? Open a PR. Include the provider, endpoint, rate limits (link to their docs), and a few notable models. Trial credits and time-limited promos don't count.

Footnotes

1. The Gemini API free tier is available to developers in the EU, UK, and Switzerland; the available regions page lists these regions. The terms still require you to use only Paid Services when you make an API Client available to users in the European Economic Area, Switzerland, or the UK. Google no longer publishes per-model free-tier rate limits; check your quotas in AI Studio. Free-tier prompts may be used by Google to improve products, except for users in the EEA, Switzerland and the UK, where the paid-services data terms also govern the unpaid quota, so those prompts are not used to improve Google products. gemini-3.1-flash-lite is on the deprecation schedule, with a shutdown date of May 7, 2027 and gemini-3.5-flash-lite as its replacement; the row stays because the model is live and free today. ↩

2. Mistral plans are global: the monthly allowance is shared across Studio, the API, and Vibe Code, so CLI usage eats the same budget (subscriptions). Free mode is the default for new accounts and needs no credit card (quickstart), and the Free plan card on the pricing page is what carries the $10/month in API credits figure quoted in the description. Free-mode inputs and outputs may be used to train Mistral models, and you can opt out at any time (data usage). Mistral no longer publishes numeric free-tier rate limits and points you at the Limits page of the admin panel instead (rate limits); the rate limit column is kept from the last published values. ↩

3. Registration accepts overseas phone numbers (registration FAQ) and the chat API does not require real-name verification: 目前调用 API 并不强制要求实名认证 (authentication FAQ). The Batch API does require it (batch FAQ). The same free models are served from the international platform at https://api.z.ai/api/paas/v4 (endpoint), where GLM-4.7-Flash, GLM-4.5-Flash and GLM-4.6V-Flash are all priced Free (pricing). Z AI has announced that GLM-4.5-Flash will be retired and its requests auto-routed to GLM-4.7-Flash (model page); the announced date has already passed while the model is still catalogued and still priced Free, so treat that row as living on borrowed time. ↩

4. The 10,000 free Neurons are shared across all Workers AI usage, not per model, and all limits reset daily at 00:00 UTC. Going over does not bill you, the request fails. Five models are excluded from Workers Free billing and need the Workers Paid plan or prepaid AI Gateway credits: @cf/moonshotai/kimi-k2.6, @cf/moonshotai/kimi-k2.7-code, @cf/zai-org/glm-5.2, @cf/deepseek-ai/deepseek-v4-flash-0731, @cf/deepseek-ai/deepseek-v4-pro-0813 (pricing). ↩

5. Groq shut down llama-3.3-70b-versatile and llama-3.1-8b-instant on August 16, 2026 (deprecations). Remaining free-plan limits vary by model: compound and compound-mini get 250 RPD, most others 1,000 RPD (rate limits). ↩

6. Kilo Code's free pool changes frequently, and the /api/gateway/models catalog can lag what is actually served: probe results have confirmed models absent from the catalog still answering. Every row listed here answered a live request between 2026-08-19 and 2026-08-21. Free models are reachable with no API key, at 200 requests per hour per IP (authentication). The kilo-auto/free router picks a model from the free pool, and Kilo's docs warn it "may route your requests to providers that log prompts and outputs". The NVIDIA free endpoints carry NVIDIA's own condition, quoted on Kilo's models page: "Trial use only - do not submit personal or confidential data. Your use is logged for security purposes and to improve NVIDIA products and services." ↩

7. LLM7.io rotates its catalog frequently, so the model list changes between checks. Access is tier-based, not token-based: turbo models are reachable anonymously or with a free token, pro models need the paid plan (models API). Anonymous access is capped at 1 RPS, 10 RPM, 60 requests per hour and 500,000 tokens per 24 hours; a free token raises that to 2 RPS, 40 RPM, 100 requests per hour and 1,000,000 tokens per 24 hours (limits). A paid Pro plan is available at $12/month. Anonymous access with no key was confirmed by live request on 2026-08-21. ↩

8. API-Inference is free for registered users. Current published limits are 2,000 requests/day per user (total across models), with per-model daily quotas dynamically adjusted and capped at 500; concurrency is also dynamically rate-limited. Requires Alibaba Cloud account binding and real-name verification (limits, intro). ↩

9. Ollama Cloud measures usage by input, cached input, and output tokens weighted per model (FAQ). Free tier has session limits resetting every 5 hours and weekly limits resetting every 7 days. Cloud models are also served through Ollama's OpenAI-compatible endpoint at ollama.com/v1. ↩

10. Free models default to 50 RPD per model. A one-time purchase of $10+ in credits unlocks 1,000 RPD for free models. OpenRouter also offers a Free Models Router (openrouter/free) and model fallbacks for chaining models in priority order. Free providers may log prompts for training. ↩

11. OVHcloud AI Endpoints offers a permanent free anonymous tier (2 requests per minute per IP, per model) with no signup or API key required. Higher rate limits (400 RPM per Public Cloud project per model) require an API key and are billed pay-as-you-go per token; new Public Cloud accounts get up to $200 in free trial credits. Models are hosted in EU data centers. ↩

12. SiliconFlow requires real-name identity verification to use free models (effective May 15, 2026, per the release notes). Verification supports mainland-Chinese documents; international users must contact support. ↩