Developer Tools1 day ago
Self-operated inference at the market floor on TokenGO
OpenAI-compatible API for running open-weight LLMs and video models.
bingus-bongo · HN
Use GLM-5.3 in Cursor today via tokengo API
The full gallery
Tech stack
19 projects
OpenAI-compatible API for running open-weight LLMs and video models.
bingus-bongo · HN
Use GLM-5.3 in Cursor today via tokengo API
Unified API gateway for auditing token consumption and analyzing AI API costs across multiple providers.
@amiuchat · X
我做了一个桌面工具：Token Switch。 给 Codex / Claude Code / OpenCode 重度用户统一管理 Provider、模型、Token、Base URL、连通测试和消耗统计。 多个 Agent，一个 Token 工作台。欢迎试用，也欢迎吐槽你最烦的配置问题。
Inject engineered cognitive abilities into AI agents at inference time.
@frank_brsrk · X
reasoning tools for ai agents
Verify AI agent decisions locally with transparent reasoning and MCP integration.
brandon_grutkowski · Product Hunt
Aurora Glass-box Quantitative AI for Humans and Agents
Buy and sell AI inference token capacity on a non-custodial spot market.
royashbrook · HN
Show HN: Mtok.market – a non-custodial spot market for AI inference tokens
Deploy AI inference models on serverless GPUs with sub-200ms cold starts and pay-per-second billing.
@svpino · X
You can check out Runpod here: Thanks to the Runpod team for partnering with me on this post.
8 AIs spent thousands of tokens debating investing philosophies. This tool turns their consensus into your personal AI investment coaching system — with a Prompt Library you can us
love0972 · HN
Which investing school are you? Free AI diagnostic and Prompt Library
Add persistent memory and knowledge to AI agents with drop-in files, URLs, and native MCP integration.
kitforai · GitHub
kitforai Kit for AI developer hub — official SDK, Claude Code plugin, MCP setup, and llms.txt.
Gateway for AI agents that prevents prompt injection, scans for secrets, and saves tokens.
benjamin_jorgensen1 · Product Hunt
Constellation Gate AI Prompt injection and token savings - #1 in benchmarks
AI toolkit for generating debate replies and analyzing arguments with seven specialized modes.
@recurno · X
Just shipped Riposte. Paste a Reddit dunk aimed at you → get 4 reply angles that take your side. Sharp. Logical. Aggressive. Socratic. Or spar the AI in Arena first. #debate #buildinpublic
Compare and evaluate AI models across coding, reasoning, agents, and other benchmarks.
davidtsong · HN
Benchmarklist: track AI benchmarks (2.4k+), models, and capabilities
Access top-tier AI models through a unified API with anonymous, borderless payments.
@xiaoheihei257 · X
📢 大消息！Gemini 3.6 Flash 和 Gemini 3.5 Flash-Lite 已经在 API 正式上线了！🚀 最近模型更新越来越频繁，这次 直接把两个新版本推出来，实际用起来感觉又进了一步。 先说 Gemini 3.6 Flash： 它是 3.5 Flash 的升级版，输出质量明显更好，但价格完全没变。最关键的是 token 消耗减少了大概 17%，遇到 DeepSWE 这种复杂代码生成任务，最多能省下 65% 的 token，成本直接降下来了，性价比很高。 再看 Gemini 3.5 Flash-Lite： 这是 3.5 系列里跑得最快、最省钱的那个，输出速度最高能达到 350 tokens/s。特别适合做高频任务，比如批量处理文档、Agent 实时搜索这些场景，用起来又快又稳，不会卡顿。 的模型生态现在越来越丰富了，两个新模型都支持官方 API 直接调用，开发者用着也方便。整体看下来，AI Agent 时代的底层支持又扎实了不少。 有在做 AI 项目或者日常调用大模型的朋友，不妨去试试新版本，体验应该会挺惊喜的～ @justinsuntron @BAI_AGI #TRONEcoStar