# Free LLM APIs

> 66 free LLM models and offers from 13 API providers with their published rate limits (requests and tokens per minute and per day), whether a card is needed, and what the provider does with free-tier data. Every value comes from the provider's official page, checked 2026-10-25; unpublished values are null.

Machine-readable data: https://free.branolabs.com/api/free-tiers.json

## Providers
- Google Gemini API: Free tier for an active Google Cloud project or free trial, no billing needed. Many Gemini models are free of charge on the free tier. Limits: Limits are set per model and per project and are not published on the docs page; Google says to view your active limits in Google AI Studio. Requests per day reset at midnight Pacific time.. Card needed: no. Source: https://ai.google.dev/gemini-api/docs/pricing (checked 2026-10-25). Details: https://free.branolabs.com/provider/google-gemini
- Groq: Free plan with published per-model limits for requests and tokens per minute and per day. Limits: Limits apply at the organization level. Cached tokens do not count towards limits.. Card needed: not stated. Source: https://console.groq.com/docs/rate-limits (checked 2026-10-25). Details: https://free.branolabs.com/provider/groq
- Cerebras Inference: Free tier for development, evaluation and experimentation (not production), with published limits. Limits: Free tier: 5 requests/min, 30K uncached tokens/min, 90K total tokens/min, 1M tokens/hour and 1M tokens/day per model.. Card needed: not stated. Source: https://inference-docs.cerebras.ai/support/rate-limits (checked 2026-10-25). Details: https://free.branolabs.com/provider/cerebras
- SambaNova Cloud: Free tier applies when no payment method is linked to the account. Limits: Free tier: 20 requests/min, 20 requests/day and 200K tokens/day per model.. Card needed: no. Source: https://docs.sambanova.ai/docs/en/models/rate-limits (checked 2026-10-25). Details: https://free.branolabs.com/provider/sambanova
- Mistral AI (La Plateforme): Free mode: API access is enabled by default with no credit card required, with included monthly usage and rate limits. Limits: The numbers are shown on the Limits page in Mistral's console, not in the public docs.. Card needed: no. Source: https://docs.mistral.ai/getting-started/quickstarts/studio/activate-and-generate-api-key (checked 2026-10-25). Details: https://free.branolabs.com/provider/mistral
- OpenRouter: Free model variants (IDs ending in ":free") can be used without buying credits, with lower daily limits until you have bought at least 10 credits. Limits: Free variants: 20 requests/min; 50 requests/day if you bought less than 10 credits in total, 1,000 requests/day with at least 10. A negative balance blocks free models too.. Card needed: no. Source: https://openrouter.ai/docs/api-reference/limits (checked 2026-10-25). Details: https://free.branolabs.com/provider/openrouter
- OpenCode Zen: Several models are free for a limited time. Free-tier chat models only work inside the OpenCode client, not from other tools. Limits: not published on the public page. Card needed: not stated. Source: https://opencode.ai/docs/zen/ (checked 2026-10-25). Details: https://free.branolabs.com/provider/opencode-zen
- Z.ai (GLM): Three Flash models are listed as free for input, cached input and output on Z.ai's pricing page. Limits: not published on the public page. Card needed: not stated. Source: https://docs.z.ai/guides/overview/pricing (checked 2026-10-25). Details: https://free.branolabs.com/provider/zai
- Cloudflare Workers AI: 10,000 Neurons per day free on both the Workers Free and Paid plans; usage beyond that costs $0.011 per 1,000 Neurons on the Paid plan. Limits: 10,000 Neurons/day, reset at 00:00 UTC. How many tokens that buys depends on the model's Neuron rate (for example llama-3.2-1b: 2,457 Neurons per million input tokens).. Card needed: not stated. Source: https://developers.cloudflare.com/workers-ai/platform/pricing/ (checked 2026-10-25). Details: https://free.branolabs.com/provider/cloudflare-workers-ai
- Vercel AI Gateway: Every Vercel team gets $5 of AI Gateway credit per month for Free Tier eligible models, rate-limited per model. Limits: $5/month of credit; free-tier requests have lower per-model rate limits than the paid tier (numbers not published on the pricing page). Buying credits moves the team to the paid tier and ends the monthly free credit.. Card needed: not stated. Source: https://vercel.com/docs/ai-gateway/pricing (checked 2026-10-25). Details: https://free.branolabs.com/provider/vercel-ai-gateway
- Scaleway Generative APIs: Free tier to try new models: 1 million tokens and 60 minutes of audio transcription. Limits: 1M tokens and 60 minutes of audio transcription. The page does not say whether this resets.. Card needed: not stated. Source: https://www.scaleway.com/en/pricing/model-as-a-service/ (checked 2026-10-25). Details: https://free.branolabs.com/provider/scaleway
- Cohere: Free trial (evaluation) API keys for testing, limited to 1,000 API calls per month. Limits: Trial keys: 1,000 API calls/month in total; chat models 20 requests/min, Embed 2,000 inputs/min, Rerank 10 requests/min, audio transcription 5 requests/min.. Card needed: not stated. Source: https://docs.cohere.com/docs/rate-limits (checked 2026-10-25). Details: https://free.branolabs.com/provider/cohere
- NVIDIA NIM API (build.nvidia.com): Members of the free NVIDIA Developer Program get free access to NIM API endpoints for prototyping. Limits: not published on the public page. Card needed: not stated. Source: https://docs.api.nvidia.com/nim/docs/product (checked 2026-10-25). Details: https://free.branolabs.com/provider/nvidia-nim

## Free models with published numbers
- openai/gpt-oss-120b (Groq): requests 30/min · 1K/day; tokens 8K/min · 200K/day
- openai/gpt-oss-20b (Groq): requests 30/min · 1K/day; tokens 8K/min · 200K/day
- openai/gpt-oss-safeguard-20b (Groq): requests 3/min · 1K/day; tokens 2K/min · 200K/day
- qwen/qwen3.8-27b (Groq): requests 30/min · 1K/day; tokens 8K/min · 200K/day
- whisper-large-v3 (Groq): requests 20/min · 2K/day; tokens n/p
- whisper-large-v3-turbo (Groq): requests 20/min · 2K/day; tokens n/p
- meta-llama/llama-prompt-guard-2-22m (Groq): requests 30/min · 14.4K/day; tokens 15K/min · 500K/day
- meta-llama/llama-prompt-guard-2-86m (Groq): requests 30/min · 14.4K/day; tokens 15K/min · 500K/day
- canopylabs/orpheus-v1-english (Groq): requests 10/min · 100/day; tokens 1.2K/min · 3.6K/day
- canopylabs/orpheus-arabic-saudi (Groq): requests 10/min · 100/day; tokens 1.2K/min · 3.6K/day
- gpt-oss-120b (Cerebras): requests 5/min; tokens 30K/min · 1M/day
- qwen-3.8-27b (Cerebras): requests 5/min; tokens 30K/min · 1M/day
- DeepSeek-V3.1 (SambaNova): requests 20/min · 20/day; tokens 200K/day
- Meta-Llama-3.3-70B-Instruct (SambaNova): requests 20/min · 20/day; tokens 200K/day
- gpt-oss-120b (SambaNova): requests 20/min · 20/day; tokens 200K/day
- DeepSeek-V3.2 (SambaNova): requests 20/min · 20/day; tokens 200K/day
- gemma-4-31B-it (SambaNova): requests 20/min · 20/day; tokens 200K/day
- cohere/north-mini-code:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- dots-studio/dots-3-note-preview:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- google/gemma-4-26b-a4b-it:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- google/gemma-4-31b-it:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- liquid/lfm-2.5-2.6b:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- nvidia/nemotron-3-super-120b-a12b:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- nvidia/nemotron-3-ultra-550b-a55b:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- nvidia/nemotron-3.5-content-safety:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- nvidia/nemotron-3.5-lightning:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- nvidia/nemotron-3.5-super-vl:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- poolside/laguna-s-2.1:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- poolside/laguna-xs-2.1:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- recursive/flint-v1.0:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- thinkingmachines/inkling-small:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- thinkingmachines/inkling:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- xingchen-agi/xing4.0-29b-a4b:free (OpenRouter): requests 20/min · 50/day; tokens n/p
- command-models (Cohere): requests 20/min; tokens n/p
- rerank (Cohere): requests 10/min; tokens n/p
- transcribe (Cohere): requests 5/min; tokens n/p

## Gone
- GitHub Models: Retired. Fully retired on July 30, 2026: playground, catalog, inference API and BYOK are no longer available. Source: https://docs.github.com/en/github-models/use-github-models/prototyping-with-ai-models
- Hugging Face Inference Providers: No free credits for free accounts. Free users get no monthly credits and must buy credits; PRO users and Team/Enterprise seats get $2.00 per month. Source: https://huggingface.co/docs/inference-providers/pricing

## Pages
- [Sources](https://free.branolabs.com/sources)
- [Methodology](https://free.branolabs.com/about)

## Optional
- [LLM API price tracker](https://prices.branolabs.com): paid per-token prices across providers
- [@BranoLabs on X](https://x.com/BranoLabs)
- [BranoLabs](https://branolabs.com): more small, data-backed AI tools