# Free LLM API

> 26 API providers that publish a free tier for large language models, 24 of them without a credit card, plus setup instructions for 5 coding agents. https://xyzs996.github.io/free-llm-api/

Every limit below is the number the provider prints on its own page, with the date that page was read — not a benchmark and not an estimate. A provider that publishes no number is recorded as publishing none; the field is never filled in with a guess. Newest source read: 2026-08-22.

## Providers

- [Google Gemini API](https://xyzs996.github.io/free-llm-api/provider/gemini.html): free tier · limits depend on the account tier · no credit card · OpenAI-compatible at https://generativelanguage.googleapis.com/v1beta/openai/ · sources read 2026-08-22.
- [GroqCloud](https://xyzs996.github.io/free-llm-api/provider/groq.html): free tier · 30 requests/minute, 1000 requests/day · no credit card · OpenAI-compatible at https://api.groq.com/openai/v1 · sources read 2026-08-22.
- [SambaNova Cloud](https://xyzs996.github.io/free-llm-api/provider/sambanova.html): free tier · 20 requests/minute, 20 requests/day · no credit card · OpenAI-compatible at https://api.sambanova.ai/v1 · sources read 2026-08-22.
- [Cohere](https://xyzs996.github.io/free-llm-api/provider/cohere.html): free tier · 20 requests/minute · no credit card · OpenAI-compatible at https://api.cohere.ai/compatibility/v1 · sources read 2026-08-22.
- [Cloudflare Workers AI](https://xyzs996.github.io/free-llm-api/provider/cloudflare-workers-ai.html): free tier · limits documented in compute units · no credit card · OpenAI-compatible at https://api.cloudflare.com/client/v4/accounts/ACCOUNT_ID/ai/v1 · sources read 2026-08-22.
- [Hugging Face Inference Providers](https://xyzs996.github.io/free-llm-api/provider/huggingface.html): free tier · limits documented as a credit balance · no credit card · OpenAI-compatible at https://router.huggingface.co/v1 · sources read 2026-08-22.
- [SiliconFlow](https://xyzs996.github.io/free-llm-api/provider/siliconflow.html): free tier · 1000 requests/minute · no credit card · OpenAI-compatible at https://api.siliconflow.com/v1 · sources read 2026-08-22.
- [Fireworks AI](https://xyzs996.github.io/free-llm-api/provider/fireworks.html): free tier · 10 requests/minute · no credit card · OpenAI-compatible at https://api.fireworks.ai/inference/v1 · sources read 2026-08-22.
- [Z.AI Open Platform](https://xyzs996.github.io/free-llm-api/provider/zai.html): free tier · free models listed rather than rate-limited · no credit card · OpenAI-compatible at https://api.z.ai/api/paas/v4 · sources read 2026-08-22.
- [Novita AI](https://xyzs996.github.io/free-llm-api/provider/novita.html): metered access · no limits published · no credit card · OpenAI-compatible at https://api.novita.ai/openai · sources read 2026-08-22.
- [Mistral La Plateforme](https://xyzs996.github.io/free-llm-api/provider/mistral.html): free tier · no limits published · no credit card · OpenAI-compatible at https://api.mistral.ai/v1 · sources read 2026-08-22.
- [Alibaba Cloud Model Studio](https://xyzs996.github.io/free-llm-api/provider/dashscope.html): free tier · limits documented per model · no credit card · OpenAI-compatible at https://dashscope-intl.aliyuncs.com/compatible-mode/v1 · sources read 2026-08-22.
- [Moonshot AI (Kimi)](https://xyzs996.github.io/free-llm-api/provider/moonshot.html): metered access · limits documented per tier · no credit card · OpenAI-compatible at https://api.moonshot.ai/v1 · sources read 2026-08-22.
- [Pollinations.AI](https://xyzs996.github.io/free-llm-api/provider/pollinations.html): free tier · 4 requests/minute · no credit card · OpenAI-compatible at https://text.pollinations.ai/openai · sources read 2026-08-22.
- [Ollama Cloud](https://xyzs996.github.io/free-llm-api/provider/ollama-cloud.html): free tier · no limits published · no credit card · OpenAI-compatible at https://ollama.com/v1 · sources read 2026-08-22.
- [Cerebras Inference](https://xyzs996.github.io/free-llm-api/provider/cerebras.html): trial credit · 5 requests/minute · credit card required · OpenAI-compatible at https://api.cerebras.ai/v1 · sources read 2026-08-22.
- [Vercel AI Gateway](https://xyzs996.github.io/free-llm-api/provider/vercel-ai-gateway.html): trial credit · limits documented as a credit balance · no credit card · OpenAI-compatible at https://ai-gateway.vercel.sh/v1 · sources read 2026-08-22.
- [IBM watsonx.ai](https://xyzs996.github.io/free-llm-api/provider/watsonx.html): trial credit · no limits published · no credit card · not OpenAI-compatible · sources read 2026-07-25.
- [OpenRouter](https://xyzs996.github.io/free-llm-api/provider/openrouter.html): free model aggregator · 20 requests/minute, 50 requests/day · no credit card · OpenAI-compatible at https://openrouter.ai/api/v1 · sources read 2026-08-22.
- [GitHub Models](https://xyzs996.github.io/free-llm-api/): free tier retired · shut down, no endpoint left to call · no credit card · OpenAI-compatible at https://models.github.ai/inference · sources read 2026-08-22 · retired 2026-07-30 · in the catalog table, no page of its own.
- [Together AI](https://xyzs996.github.io/free-llm-api/provider/together.html): metered access · no fixed numbers published · no credit card · OpenAI-compatible at https://api.together.xyz/v1 · sources read 2026-08-22.
- [Nebius Token Factory](https://xyzs996.github.io/free-llm-api/provider/nebius.html): metered access · no fixed numbers published · no credit card · OpenAI-compatible at https://api.tokenfactory.nebius.com/v1 · sources read 2026-08-22.
- [Perplexity API](https://xyzs996.github.io/free-llm-api/provider/perplexity.html): metered access · 50 requests/minute · no credit card · OpenAI-compatible at https://api.perplexity.ai · sources read 2026-08-22.
- [DeepInfra](https://xyzs996.github.io/free-llm-api/provider/deepinfra.html): metered access · no limits published · no credit card · OpenAI-compatible at https://api.deepinfra.com/v1/openai · sources read 2026-08-22.
- [Chutes](https://xyzs996.github.io/free-llm-api/provider/chutes.html): metered access · limits documented per plan · no credit card · OpenAI-compatible at https://llm.chutes.ai/v1 · sources read 2026-08-22.
- [Scaleway Generative APIs](https://xyzs996.github.io/free-llm-api/provider/scaleway.html): metered access · no limits published · credit card required · OpenAI-compatible at https://api.scaleway.ai/v1 · sources read 2026-07-25.

## What changed, as of 2026-08-22

Every source above was read again on this date. What moved since the previous read:

- **GitHub Models** (lifecycle): Retired on 2026-07-30 as announced: playground, model catalog, inference API and BYOK endpoints are all gone. Kept as a tombstone entry.
- **Novita AI** (lifecycle): inclusionai/Ling-3.0-flash and Mind Lab Macaron V1 Venti are no longer priced at zero; no model on the pricing table is free any more. Moved to metered access.
- **GroqCloud** (limit-changed): llama-3.3-70b-versatile and llama-3.1-8b-instant left the free plan table. openai/gpt-oss-safeguard-20b, qwen/qwen3.6-27b and groq/compound-mini are in it. The RPM/RPD columns now quote openai/gpt-oss-120b at 30 RPM / 1K RPD.
- **Cerebras Inference** (limit-changed): zai-glm-4.7 left the Free Trial table; only gpt-oss-120b and gemma-4-31b remain at 5 RPM / 30K TPM / 1M TPH / 1M TPD.
- **Z.AI Open Platform** (added): GLM-4.6V-Flash, a vision model, is now listed at $0 across every pricing column alongside GLM-4.7-Flash and GLM-4.5-Flash.
- **SambaNova Cloud** (added): DeepSeek-V3.2 and gemma-4-31B-it appear in the free table as preview models, on the same 20 RPM / 20 RPD / 200K TPD numbers.
- **Pollinations.AI** (correction): The API docs do publish limits after all, as a minimum interval per tier: Anonymous one request every 15s, Seed 5s, Flower 3s, Nectar none. The entry previously said no figures were published.
- **Nebius Token Factory** (correction): The 60 RPM / 400,000 TPM pair carried here as a baseline appears on the source page only inside a table labelled a visual example. No account default is published; the entry no longer quotes a number.
- **Vercel AI Gateway** (correction): The $5 monthly free credit is no longer published on the pricing page. The free tier still exists but no amount is stated, so none is quoted here.
- **Cloudflare Workers AI** (correction): Some models sit outside the 10,000 Neuron allowance and need a paid billing method regardless: @cf/moonshotai/kimi-k2.6, @cf/zai-org/glm-5.2 and @cf/deepseek-ai/deepseek-v4-pro-0813.
- **Moonshot AI (Kimi)** (lifecycle): Docs moved to platform.kimi.ai. The lowest tier still requires a $1 recharge before any call goes through, so the entry moved out of the permanent free tier list into metered access.

First re-check since publication. Every source was read again: GitHub Models is gone, Novita has no free rows left, and Groq dropped both Llama models from the free table. Two entries could not be re-read and keep their old check date.

- [changelog.json](https://github.com/xyzs996/free-llm-api/blob/main/data/changelog.json): every week on record, same fields.

## Model families

Which providers in the catalog serve a given family, and what each one limits it to.

- [gpt-oss](https://xyzs996.github.io/free-llm-api/model/gpt-oss.html): OpenAI. Offered by 10 of the 26 providers here.
- [Llama](https://xyzs996.github.io/free-llm-api/model/llama.html): Meta. Offered by 9 of the 26 providers here.
- [Qwen](https://xyzs996.github.io/free-llm-api/model/qwen.html): Alibaba. Offered by 10 of the 26 providers here.
- [DeepSeek](https://xyzs996.github.io/free-llm-api/model/deepseek.html): DeepSeek. Offered by 5 of the 26 providers here.
- [Mistral](https://xyzs996.github.io/free-llm-api/model/mistral.html): Mistral AI. Offered by 4 of the 26 providers here.
- [GLM](https://xyzs996.github.io/free-llm-api/model/glm.html): Z.ai. Offered by 3 of the 26 providers here.
- [Kimi](https://xyzs996.github.io/free-llm-api/model/kimi.html): Moonshot AI. Offered by 2 of the 26 providers here.

## Coding agents

Which of these providers a given tool can be pointed at, and the exact configuration.

- [Codex CLI](https://xyzs996.github.io/free-llm-api/client/codex.html)
- [Claude Code](https://xyzs996.github.io/free-llm-api/client/claude-code.html)
- [Continue](https://xyzs996.github.io/free-llm-api/client/continue.html)
- [Cursor](https://xyzs996.github.io/free-llm-api/client/cursor.html)
- [Cline](https://xyzs996.github.io/free-llm-api/client/cline.html)

## Machine-readable

- [providers.json](https://xyzs996.github.io/free-llm-api/providers.json): every field above, including the model lists and the official source URLs each limit was read from.
- [sitemap.xml](https://xyzs996.github.io/free-llm-api/sitemap.xml): every page, with the date its own sources were last read.
- [Key checker](https://xyzs996.github.io/free-llm-api/verify.html): checks a key against a provider from the browser; the key goes to that provider and nowhere else.

## What it costs once a free tier ends

Out of scope for this catalog, which stops where the free tier does. A sibling project re-reads OpenRouter's whole catalog every day. On **2026-08-22** it held 60 models with a list price; these were the cheapest by input price:

- **Gemini 3.7 Flash** — $0.1875 per million input tokens, $0.9375 per million output (batch/queued price, not the interactive one), ranked #3 in the Design Arena `androidnative` agent category.
- **Gemini 3 Flash Preview** — $0.25 per million input tokens, $1.50 per million output (batch/queued price, not the interactive one), ranked #9 in the Design Arena `agenticslides` agent category.
- **MiniMax M3** — $0.30 per million input tokens, $1.20 per million output, ranked #10 in the Design Arena `python-pptxslides` agent category.
- **Gemini 3.6 Flash** — $0.375 per million input tokens, $1.875 per million output (batch/queued price, not the interactive one), ranked #6 in the Design Arena `agenticgamedev` agent category.
- **GLM 4.7** — $0.40 per million input tokens, $1.75 per million output, ranked #27 in the Design Arena `androidnative` agent category.
- Read on 2026-08-22 from OpenRouter's public catalog. All 60 rows: [JSON](https://cdn.jsdelivr.net/gh/xyzs996/llm-api-pricing@main/data/prices.json) · [CSV](https://cdn.jsdelivr.net/gh/xyzs996/llm-api-pricing@main/data/prices.csv) · [repository](https://github.com/xyzs996/llm-api-pricing).

A list price is what a vendor publishes. What a month of it came to is a different number, and the same project keeps those separately — every figure it has cited, with the sentence it appeared in. The model and coding-agent pages here quote the ones that name their own subject:

- `$0.06 / $0.2` per million — "At the low end, MiniMax M3 runs $0.06 to $0.2 per million and draws 60% to 70% of its revenue from outside its home market." ([1.6 Billion Free Tokens Is a Compression Ratio, Not a Strategy](https://github.com/xyzs996/llm-api-pricing/discussions/12))
- `$0.19 / $5` per million tokens — "Chinese AI models provide a cost-effective alternative to their American counterparts, with input costs as low as $0.19 per million tokens, compared to OpenAI's $5-12." ([How Chinese AI Agent Tools Leverage 1.6 Billion Free Tokens](https://markyanai.medium.com/how-chinese-ai-agent-tools-leverage-1-6-billion-free-tokens-69b483c4eb6a))
- `$1` per million tokens — "Top-tier Chinese models such as GLM5.2 and DeepSeek V4 Pro sit near $1 per million tokens at inference gross margins of 10% to 20%." ([1.6 Billion Free Tokens Is a Compression Ratio, Not a Strategy](https://github.com/xyzs996/llm-api-pricing/discussions/12))
- `$1.25 / $4.25` per million — "Meta priced Muse Spark 1.1 at $1.25 per million input and $4.25 per million output, roughly 75% and 83% below Anthropic's Opus, and the tradeoff is visible in the benchmarks, since it leads on MCP Atlas and JobBench while trailing on SWE-Bench Pro and DeepSWE 1.1." ([1.6 Billion Free Tokens Is a Compression Ratio, Not a Strategy](https://github.com/xyzs996/llm-api-pricing/discussions/12))
- `$3` per million input tokens — "The $3 per million input tokens price point means developers should carefully evaluate whether the premium model's capabilities justify the increased costs for their specific use cases." ([Choosing the Right AI Model for Coding: Cost vs. Efficiency](https://github.com/xyzs996/llm-api-pricing/discussions/22))
- [The whole table](https://xyzs996.github.io/llm-api-pricing/figures.html): all figures, not only prices.

## Questions answered in full, with the figures behind them

- **Which free LLM APIs work without a credit card — and what are the published limits?** — every permanent free tier that asks for no card, with the limit each one publishes and the ones that publish none. https://github.com/xyzs996/free-llm-api/discussions/2
- **What are the Gemini, Groq and OpenRouter free tier rate limits right now?** — the per-minute and per-day figures each one documents, why one of the three publishes no single number, and what that means for quoting it. https://github.com/xyzs996/free-llm-api/discussions/3
- **Which free LLM APIs are OpenAI compatible, and what base URL do I point my coding agent at?** — the base URL for every compatible provider, the one that is not a drop-in, and what the compatible flag does not promise. https://github.com/xyzs996/free-llm-api/discussions/4

Three more this catalog is asked and does not answer, because they start where its free tiers end. Each is answered by a figure quoted above, in the sentence it was published in:

- **Are Chinese models actually cheaper than OpenAI per million tokens, and by how much?** — "Chinese AI models provide a cost-effective alternative to their American counterparts, with input costs as low as $0.19 per million tokens, compared to OpenAI's $5-12." ([How Chinese AI Agent Tools Leverage 1.6 Billion Free Tokens](https://markyanai.medium.com/how-chinese-ai-agent-tools-leverage-1-6-billion-free-tokens-69b483c4eb6a))
- **What gross margin sits behind a $1-per-million-token price?** — "Top-tier Chinese models such as GLM5.2 and DeepSeek V4 Pro sit near $1 per million tokens at inference gross margins of 10% to 20%." ([1.6 Billion Free Tokens Is a Compression Ratio, Not a Strategy](https://github.com/xyzs996/llm-api-pricing/discussions/12))
- **Is a premium coding model worth $3 per million input tokens?** — "The $3 per million input tokens price point means developers should carefully evaluate whether the premium model's capabilities justify the increased costs for their specific use cases." ([Choosing the Right AI Model for Coding: Cost vs. Efficiency](https://github.com/xyzs996/llm-api-pricing/discussions/22))

## Elsewhere

Three sibling sites publish their own `llms.txt` in this same shape — one request, every address, each one dated:

- [https://xyzs996.github.io/llm-api-pricing/llms.txt](https://xyzs996.github.io/llm-api-pricing/llms.txt): every price figure it has published, quoted inline with the sentence and the date it was published in, plus the write-ups behind them.
- [https://xyzs996.github.io/free-proxy-health-list/llms.txt](https://xyzs996.github.io/free-proxy-health-list/llms.txt): free HTTP/SOCKS proxies, re-checked and re-published on a schedule, as TXT, JSON and CSV.
- [https://xyzs996.github.io/iptv-doctor/llms.txt](https://xyzs996.github.io/iptv-doctor/llms.txt): which public IPTV channels answer right now, per country and per channel. No stream URL is published.

## Corrections

A limit here is only as good as the day it was read. If one has moved, or a provider is missing, [one line in this form](https://github.com/xyzs996/free-llm-api/issues/new?template=heard.yml&came_from=llms.txt) takes a plain sentence and no evidence — a single field; the correction form asks for the provider's page and the date you read it.

This catalog is also published in [简体中文](https://xyzs996.github.io/free-llm-api/zh/), page for page.