[2026-11-04] auto-update via provider APIs: 6 models added, 1 removed. ↓ history

Free Inference

[ 57 models ][ 9 providers ][ checked daily · last change 2026-11-04 ] ⟶

Every model you can run for free right now — limits, expiry and catches included. One table, no accounts. Click a model to get it: its provider page, key and API.

01 — Models

Model Type Released Provider Context Free limits Until Catch Score
stealth/crouton-alphaomni2026-1103OpenRouter (:free)1.04858M20 rpm · 50 req/dayongoing429strains■■■□□
nvidia/nemotron-3.5-super-vl:freevision2026-1003OpenRouter (:free)1M20 rpm · 50 req/dayongoing429strains■■■□□
xingchen-agi/xing4.0-29b-a4b:free2026-1003OpenRouter (:free)256k20 rpm · 50 req/dayongoing429strains■■□□□
recursive/flint-v1.0:free2026-1003OpenRouter (:free)256k20 rpm · 50 req/dayongoing429strains■■□□□
nemotron-3.5-super-vl-120b-a12bvision2026-1005NVIDIA NIM—unpublishedongoingeval■■□□□
mistral-large-4vision2026-1006Mistral La Plateforme—per-model (see docs)ongoing■■□□□
nvidia/nemotron-3.5-lightning-free2026-1007Vercel AI Gateway1M$5 credit / monthongoing■■■■□
reflection/beam-free2026-1007Vercel AI Gateway1.04858M$5 credit / monthongoing■■■■□
stealth/glyph-cluster2026-1007Vercel AI Gateway250k$5 credit / monthongoing■■■□□
reflection/beam2026-1007Vercel AI Gateway1.04858M$5 credit / monthongoing■■■■□
Gemini 3.8 Flashvision2026-0901Google AI Studio1M~1,500 req/day (Flash)ongoingtrains■■■■■
deepseek-v4.1-flash2026-0905NVIDIA NIM—unpublishedongoingeval■■□□□
qwen3.8-27b2026-0802Groq131k1k req/day · 8k tok/minongoingslow■■■■□
nvidia/nemotron-3.5-lightning:free2026-0803OpenRouter (:free)1M20 rpm · 50 req/dayongoing429strains■■■□□
dots-studio/dots-3-note-preview:freevision2026-0803OpenRouter (:free)500k20 rpm · 50 req/dayongoing429strains■■□□□
liquid/lfm-2.5-2.6b:free2026-0803OpenRouter (:free)64k20 rpm · 50 req/dayongoing429strains■■□□□
qwen3.8-27b2026-0804Cloudflare Workers AI—10k neurons/dayongoing■■□□□
nemotron-3.5-lightning-30b-a3b2026-0805NVIDIA NIM—unpublishedongoingeval■■□□□
glm-5.3-flash (3rd-party)2026-0806Mistral La Plateforme1Mper-model (see docs)ongoing■■■■□
qwen-3.8-27b2026-0808Cerebras64k$5 trial creditstrialcardcredits■□□□□
Gemini 3.6 Flashvision2026-0701Google AI Studio1M~1,500 req/day (Flash)ongoingtrains■■■■■
Gemini 3.5 Flash-Litevision2026-0701Google AI Studio1M~1,500 req/day (Flash)ongoingtrains■■■■■
nvidia/nemotron-3-ultra-550b-a55b:free2026-0603OpenRouter (:free)1M20 rpm · 50 req/dayongoing429strains■■■□□
cohere/north-mini-code:freecode2026-0603OpenRouter (:free)250k20 rpm · 50 req/dayongoing429strains■■□□□
nvidia/nemotron-3.5-content-safety:freevision2026-0603OpenRouter (:free)125k20 rpm · 50 req/dayongoing429strains■■□□□
nemotron-3-ultra-550b-a55b2026-0605NVIDIA NIM—unpublishedongoingeval■■□□□
north-mini-code-1.0code2026-0609Cohere256k1k calls / monthongoingeval■■□□□
Gemini 3.1 Flash-Litevision2026-0501Google AI Studio1M~1,500 req/day (Flash)ongoingtrains■■■■■
command-a-plus-05-20262026-0509Cohere128k1k calls / monthongoingeval■■□□□
google/gemma-4-31b-it:freevision2026-0403OpenRouter (:free)256k20 rpm · 50 req/dayongoing429strains■■□□□
google/gemma-4-26b-a4b-it:freevision2026-0403OpenRouter (:free)256k20 rpm · 50 req/dayongoing429strains■■□□□
nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:freeomni+reasoning2026-0403OpenRouter (:free)250k20 rpm · 50 req/dayongoing429strains■■□□□
gemma-4-26b-a4b-itvision2026-0404Cloudflare Workers AI—10k neurons/dayongoing■■□□□
google/lyria-3-pro-previewvision2026-0303OpenRouter (:free)1.04858M20 rpm · 50 req/dayongoing429strains■■■□□
google/lyria-3-clip-previewvision2026-0303OpenRouter (:free)1.04858M20 rpm · 50 req/dayongoing429strains■■■□□
mistral-small-4reasoning2026-0306Mistral La Plateforme—per-model (see docs)ongoing■■□□□
openrouter/free (auto-router)vision2026-0203OpenRouter (:free)195k20 rpm · 50 req/dayongoing429strains■■□□□
glm-4.7-flash2026-0104Cloudflare Workers AI131k10k neurons/dayongoing■■□□□
ministral-3-14b2025-1206Mistral La Plateforme—per-model (see docs)ongoing■■□□□
gpt-oss-120breasoning2025-0802Groq131k1k req/day · 8k tok/minongoingslow■■■■□
gpt-oss-20breasoning2025-0802Groq131k1k req/day · 8k tok/minongoingslow■■■■□
gpt-oss-120breasoning2025-0804Cloudflare Workers AI—10k neurons/dayongoing■■□□□
gpt-oss-20breasoning2025-0804Cloudflare Workers AI—10k neurons/dayongoing■■□□□
codestral (v25.08)code2025-0806Mistral La Plateforme—per-model (see docs)ongoing■■□□□
gpt-oss-120breasoning2025-0808Cerebras65k$5 trial creditstrialcardcredits■□□□□
command-a-reasoning-08-2025reasoning2025-0809Cohere256k1k calls / monthongoingeval■■□□□
Gemini 2.5 Provision2025-0601Google AI Studio1M~1,500 req/day (Flash)ongoingtrains■■■■■
Gemini 2.5 Flashvision2025-0601Google AI Studio1M~1,500 req/day (Flash)ongoingtrains■■■■■
llama-4-scout-17b-16e-instructvision2025-0404Cloudflare Workers AI—10k neurons/dayongoing■■□□□
qwen3-30b-a3b-fp82025-0404Cloudflare Workers AI—10k neurons/dayongoing■■□□□
mistral-small-3.1-24b-instructvision2025-0304Cloudflare Workers AI128k10k neurons/dayongoing■■□□□
command-a-03-20252025-0309Cohere256k1k calls / monthongoingeval■■□□□
deepseek-r1-distill-qwen-32breasoning2025-0104Cloudflare Workers AI—10k neurons/dayongoing■■□□□
llama-3.3-70b-instruct-fp8-fast2024-1204Cloudflare Workers AI—10k neurons/dayongoing■■□□□
command-r7b-12-20242024-1209Cohere128k1k calls / monthongoingeval■■□□□
qwen2.5-coder-32b-instructcode2024-1104Cloudflare Workers AI—10k neurons/dayongoing■■□□□
whisper-large-v3 (audio)audio2023-1102Groq—1k req/day · 8k tok/minongoingslow■■■■□
history
  • auto-update via provider APIs: 6 models added, 1 removed.
  • Model lists replaced with concrete live data: OpenRouter models API (16 ':free'), Vercel gateway API (3 zero-price), NVIDIA / Cerebras / Cohere / Mistral / Cloudflare official docs.
  • First snapshot: 9 providers tracked, all entries verified today.
  • Hugging Face Inference Providers no longer offer free monthly credits - free accounts get nothing, PRO gets $2/month.
  • GitHub Models retired: playground, model catalog, inference API and BYOK shut down entirely.
  • Cerebras free access became trial-only: $5 credits, verified card required, 30-day expiry (date approximate).
sources — what we watch for new / removed free models