posts.json

[
  {
    "slug": "use-any-model-in-claude-code-through-token-station",
    "title": "Use any model in Claude Code through Token Station",
    "summary": "Claude Code can run through Token Station with either a persistent ~/.claude/settings.json file or temporary ANTHROPIC_* environment variables. Set the base URL, token, Opus/Sonnet/Haiku model aliases, and subagent model, then verify with a one-line pong check.",
    "category": "tutorial",
    "date": "2026-06-18"
  },
  {
    "slug": "run-any-model-in-codex-through-token-station",
    "title": "Run any model in Codex through Token Station",
    "summary": "OpenAI's Codex can run any model, not just OpenAI's. Since February 2026 it requires the Responses API, so the platform you point it at must support that natively. The exact ~/.codex/config.toml for Token Station, verified against the OpenAI docs, plus smart routing.",
    "category": "tutorial",
    "date": "2026-06-18"
  },
  {
    "slug": "glm-5-2-zai-stock-jump",
    "title": "GLM-5.2 sent Z.AI's stock up 33%. Now try it free on Token Station",
    "summary": "Zhipu AI's Hong Kong-listed shares jumped as much as 48% the day GLM-5.2 shipped, its open-source 1M-context coding model. Why it moved, and how to try GLM-5.2 free with a Token Station signup.",
    "category": "research",
    "date": "2026-06-15",
    "cover": "blog/glm-5-2-zai-stock-jump-cover.png"
  },
  {
    "slug": "try-kimi-k2-7-code-in-your-coding-agent",
    "title": "Kimi K2.7 Code: Cheap Enough to Try, Maybe Good Enough to Share the Work",
    "summary": "Moonshot's new 1T open-weights coding model costs $0.95/$4 per million tokens. Pair it with your SOTA model in Claude Code, Codex, or OpenClaw and let it carry the routine work.",
    "category": "tutorial",
    "date": "2026-06-13",
    "cover": "blog/kimi-k2-7-code-cover.png"
  },
  {
    "slug": "try-claude-fable-5-in-codex-openclaw-and-pi",
    "title": "Try Before You Commit: Claude Fable 5 in Codex, OpenClaw, and Pi",
    "summary": "Anthropic's new flagship is state-of-the-art, controversial, and $10/$50 per million tokens. Experiment with it temporarily in your existing harness. No Anthropic account, just Token Station free credits.",
    "category": "tutorial",
    "date": "2026-06-12",
    "cover": "blog/claude-fable-5-cover.png"
  },
  {
    "slug": "use-grok-build-in-codex-and-claude-code",
    "title": "How to Use Grok Build in Codex and Claude Code",
    "summary": "xAI's Grok Build model is evolving fast and costs a fraction of GPT-5.5 or Claude Fable 5. Run it inside Claude Code or Codex with Token Station free credits. No xAI account needed.",
    "category": "tutorial",
    "date": "2026-06-10",
    "cover": "blog/claude-code-grok-build.png"
  },
  {
    "slug": "american-open-models-catching-up-nemotron-gpt55",
    "title": "American Open Models Are Catching Up: Nemotron-3 Ultra Leads US Open Weights",
    "summary": "NVIDIA Nemotron-3 Ultra is the most capable American open-weights model on the Artificial Analysis Intelligence Index, and it's free for a limited time via NVIDIA NIM.",
    "category": "research",
    "date": "2026-06-08",
    "cover": "blog/american-open-models-catching-up-nemotron-gpt55-cover.png"
  }
]