卧槽!Google Gemini 4 Hydrogen🤣api 免费调用1500次/d,被错过了,可以大家magpie聚合一下😁😁
今天翻 Muse 的情报空间,突然刷到一条 AI Studio 的免费额度消息。
我还以为就是网页上免费聊几句,仔细一看,居然是能接进自己工具里的 API!
Gemini 4 Hydrogen,100 万上下文,看到的免费额度大约每天 1500 次。
这就有意思了!🤯
操作也简单:
① 打开 Google AI Studio
② 创建一个 API Key
③ 在支持自定义模型的客户端里填入 gemini-4-hydrogen
④ 测试能否正常调用
平时拿来接 Agent、跑脚本、批量处理资料,比单纯在网页聊天实用多了。
注意看自己账号的实际额度,免费配额和模型权限不一定人人相同。
Muse 这情报空间真有点东西,差点漏掉这么大一块羊毛!😂
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me.
I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA.
I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation.
Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO.
Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours.
Jensen
is Mistral for $6/month worth it?
10 things to be aware of:
1. the student plan is kind of absurd value. for around €6/month you get €125 of Vibe usage + €7 of API usage. the regular plan costs 2x more, but also gives 2x the usage.
2. DeepSeek speed is one of the biggest differences. i’m seeing roughly 70-120 tok/s on OpenCode Go, while Mistral can sit much higher.
3. DeepSeek through Mistral can feel 2-3x faster. 100 tok/s vs 250-300 tok/s is a very noticeable difference in actual coding sessions.
4. when it’s healthy, i’ve seen Mistral push DeepSeek into the 200-400 tok/s range. sometimes even above 400.
5. the value multiplier is strong. on the regular plan, the effective usage can land around 11x direct API value once i account for how they price the quota.
6. cache hit rate is good, but not perfect. i’ve seen it sit in the low-to-mid 90% range, which is worse than the 97-99% i can get elsewhere.
7. the biggest weakness is reliability. i’ve hit 503s, rate limits and sessions where DeepSeek or GLM Flash just stop mid-task.
8. thinking config matters a lot. Mistral can run DeepSeek with thinking off by default depending on the setup, so i make sure reasoning is enabled before judging the model.
9. the Vibe API key is the one i care about. the Studio key has a much smaller allowance, so using the wrong one can make the plan look way worse than it actually is.
10. for pure DeepSeek usage, i currently prefer Mistral over OpenCode Go. it’s faster, the allowance is strong, and the student pricing makes the economics hard to ignore.
the only thing stopping it from being an easy recommendation is reliability.
when it works, the value is excellent.
Live board: https://t.co/L44GofxqKK
Every run executes in a @Runta runtime under the same limits, and shows up with its outcome, attempts, and token cost as soon as it finishes.
Rankings will shift as runs come in. Each harness × model pair is final after 248 runs.
The 62 tasks span software engineering (20), science and math (10), systems (8), data (7), ML (6), security (5) and 6 in other domains.
Harnesses include Claude Code, Codex, Pi, Oh My Pi, Hermes Agent, Cline, ZCode, KimiCode, DeepSeek Harness, Grok Build, Antigravity, OpenCode and Copilot.
Live board: https://t.co/L44GofxqKK
Every run executes in a @Runta runtime under the same limits, and shows up with its outcome, attempts, and token cost as soon as it finishes.
Rankings will shift as runs come in. Each harness × model pair is final after 248 runs.
The 62 tasks span software engineering (20), science and math (10), systems (8), data (7), ML (6), security (5) and 6 in other domains.
Harnesses include Claude Code, Codex, Pi, Oh My Pi, Hermes Agent, Cline, ZCode, KimiCode, DeepSeek Harness, Grok Build, Antigravity, OpenCode and Copilot.