Free Tokens

MiniMax M3.1 Flash is now free on OpenCode, with unlimited use.

MLH Developer Week offers a one-month X Premium subscription, including Grok and Cursor Pro; accounts must be registered for at least 30 days. The promotion ends on November 12.

Reports indicate that Claude is offering a one-week independent quota for Fable 5.2, allowing users to experience it for free for seven days. This article retains the original image links and stated facts, and contains no channel promotion or subscription guidance.

LG AI has officially opened the API for its K-EXAONE 2.5 model, offering free, unlimited access to users worldwide. Its capabilities rival those of DeepSeek and GLM, supporting the development of agent-based applications.

Kaggle offers 30 hours of free GPU access weekly (32GB VRAM), available to free accounts after verification. It's ideal for running open-source models or generating images.

Shusheng·Duan砚 offers free research model quotas and regular ink point rebates; Xiaohongshu's self-developed Dots API open platform allows free access to large models.

Protected link: https://404.li/cfsolve.

Tencent Cloud's Yuanbao Search API offers a long-term free quota of 2,000 calls per month, which can be increased to 6,000 by entering an invitation code before November 30. It supports both MCP and API integration, enabling Agents to perform online searches. Forum feedback indicates its quality is average and that it does not search for content from official WeChat accounts.

Channel image from Sliverkiss's Abandoned Culture Research Institute.

sensenova has launched the glm-5.3 model for free, with content mentioning #ai #freebie.

zcode gives away 300 million tokens at midnight every day, which can be used in the middle of the night, with a concurrency limit of 3.

Hermes officially provides free models, including stealth/crouton-alpha.