All times Eastern (ET)Provider pages checked daily at

Free AI model APIs

Which models you can call without paying, read daily from each company's own pages: models that are free to call, free tiers, and free daily tokens in exchange for sharing your data. Each row gives the limit as the company publishes it and when AIWire first saw it.

ModelKindLimitConditionsSinceSource
GPT-4.1 nanogpt-4.1-nanoFree for shared data10M tokens a day for Launch and Grow (2.5M for Build)—in place when tracking began (October 12, 2026)source
GPT-4o (2024-05-13)gpt-4o-2024-05-13Free for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
GPT-5 (2025-08-07)gpt-5-2025-08-07Free for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
GPT-5 mini (2025-08-07)gpt-5-mini-2025-08-07Free for shared data10M tokens a day for Launch and Grow (2.5M for Build)—in place when tracking began (October 12, 2026)source
GPT-5 nano (2025-08-07)gpt-5-nano-2025-08-07Free for shared data10M tokens a day for Launch and Grow (2.5M for Build)—in place when tracking began (October 12, 2026)source
GPT-5.6 Lunagpt-5.6-lunaFree for shared data10M tokens a day for Launch and Grow (2.5M for Build)—in place when tracking began (October 12, 2026)source
GPT-5.6 Solgpt-5.6-solFree for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
GPT-5.6 Terragpt-5.6-terraFree for shared data10M tokens a day for Launch and Grow (2.5M for Build)—in place when tracking began (October 12, 2026)source
GPT-6 Astragpt-6-astraFree for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
GPT-6 Lunagpt-6-lunaFree for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
GPT-6 Solgpt-6-solFree for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
GPT-6.1 Solgpt-6.1-solFree for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
o1 (2024-12-17)o1-2024-12-17Free for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
o3 (2025-04-16)o3-2025-04-16Free for shared data1M tokens a day for Launch and Grow (250K for Build)—in place when tracking began (October 12, 2026)source
o4 mini (2025-04-16)o4-mini-2025-04-16Free for shared data10M tokens a day for Launch and Grow (2.5M for Build)—in place when tracking began (October 12, 2026)source

Terms as of October 12, 2026: OpenAI will give 30 days notice before terminating this program. source

No first-party free API tier is documented.

ModelKindLimitConditionsSinceSource
Gemini 2.5 Flashgemini-2.5-flashFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 2.5 Flash-Litegemini-2.5-flash-liteFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 2.5 Flash Native Audio Preview (12-2025)gemini-2.5-flash-native-audio-preview-12-2025Free tierNot published—in place when tracking began (October 7, 2026)source
Gemini 2.5 Flash Preview TTSgemini-2.5-flash-preview-ttsFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 2.5 Progemini-2.5-proFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3 Flash Previewgemini-3-flash-previewFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.1 Flash Litegemini-3.1-flash-liteFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.1 Flash Live Previewgemini-3.1-flash-live-previewFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.1 Flash TTS Previewgemini-3.1-flash-tts-previewFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.5 Flashgemini-3.5-flashFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.5 Flash Litegemini-3.5-flash-liteFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.5 Live Translate Previewgemini-3.5-live-translate-previewFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.5 Transcribegemini-3.5-transcribeFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.5 Transcribe Livegemini-3.5-transcribe-liveFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.6 Flashgemini-3.6-flashFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.7 Flashgemini-3.7-flashFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.8 Flashgemini-3.8-flashFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.8 Flash Lite TTSgemini-3.8-flash-lite-ttsFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.8 Flash TTSgemini-3.8-flash-ttsFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.8 Livegemini-3.8-liveFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini 3.8 Live Extended Thinkinggemini-3.8-live-extended-thinkingFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini Robotics ER 2 Previewgemini-robotics-er-2-previewFree tierNot published—in place when tracking began (October 7, 2026)source
Gemini Robotics ER 2 Streaming Previewgemini-robotics-er-2-streaming-previewFree tierNot published—in place when tracking began (October 7, 2026)source

No first-party free API tier is documented.

No first-party free API tier is documented.

Terms as of October 12, 2026: Contributor mode: On Model API, Contributor mode runs the same models at heavily discounted token prices, in exchange for permission to use your content to train future Meta models. It lowers the barrier to entry for prototyping, testing integrations, and scaling experiments where it's acceptable for Meta to use your content to train future Meta models. source

No first-party free API tier is documented.

ModelKindLimitConditionsSinceSource
Mistral Moderation 2mistral-moderation-2603Free tierNot published—in place when tracking began (October 8, 2026)source

No first-party free API tier is documented.

ModelKindLimitConditionsSinceSource
GLM-4.5-Flashglm-4.5-flashFree modelNot published—in place when tracking began (October 7, 2026)source
GLM-4.6V-Flashglm-4.6v-flashFree modelNot published—in place when tracking began (October 7, 2026)source
GLM-4.7-Flashglm-4.7-flashFree modelNot published—in place when tracking began (October 7, 2026)source

No first-party free API tier is documented.

Terms as of October 8, 2026: Only models in the Singapore region with the service deployment scope set to International are eligible for a free quota. The free quota is valid for 90 days, starting from the date you activate Alibaba Cloud Model Studio, the model is released, or your model request is approved (whichever is later). Each model (such as qwen-plus, qwen3.6-plus, or qwen3.6-plus-2026-04-02) has its own independent free quota (typically 1,000,000 tokens). A snapshot version with a date suffix (for example, qwen3.7-max-2026-05-17) and the latest version without a date suffix (for example, qwen3.7-max) are treated as two independent models, each with its own free quota. The free quota only covers charges for real-time inference (invocations). source

No first-party free API tier is documented. build.nvidia.com is a free trial for trying models, not a free API tier, and NVIDIA doesn't publish its limits.

Where to find the numbers

Google and Mistral show free-tier rate limits only after you sign in: Gemini's in Google AI Studio, Mistral's in the Mistral console. AIWire can't read pages behind a login, so those rows say "Not published".

"In place when tracking began" means the row was already on the company's page the first time AIWire read it. Free access can end at short notice: check the company's page before you rely on it.