opencode.ai Zen 免费 API 详情

实测于 2026-08-30 00:00 +08:00（来源：https://opencode.ai/docs/zen + https://opencode.ai/zen/v1/models 无鉴权实测） 渠道 slug：opencode-zen | 详情页 URL：/free-llm-api-rankings/opencode-zen | Base URL：https://opencode.ai/zen/v1

1. 渠道介绍

opencode.ai Zen 是 OpenCode 团队为编程代理精选的聚合网关，聚合 8 个匿名/免费模型（Big Pickle 隐身模型、MiMo-V2.5、Hy3、Ling 3.0 Fin、Nemotron 3 Ultra/Lightning、Muse Spark Contributor、Laguna），限时免费（反馈收集期）。全部模型通过 OpenAI 兼容的 /v1/chat/completions（Muse Spark 用 /v1/responses）暴露。需 opencode.ai Zen 网关 Key（免费，$20 预付按量计费、余额 $5 自动充值）。60 RPM，200K 上下文。

2. 实测支持模型列表

口径：models_total = 8（opencode.ai /v1/models 无鉴权实测 + docs 交叉验证）；chat = 8；均为 anonymous/free 限时模型

匿名/免费模型（8 个）

• big-pickle — Big Pickle stealth anonymous
• mimo-v2.5-free — MiMo-V2.5 Free
• hy3-free — Hy3 Free
• ling-3.0-flash-fin-free — Ling 3.0 Flash Fin Free
• nemotron-3-ultra-free — Nemotron 3 Ultra Free (NVIDIA free endpoint)
• nemotron-3.5-lightning-free — Nemotron 3.5 Lightning Free (NVIDIA free endpoint)
• muse-spark-1.2-contributor-free — Muse Spark 1.2 Contributor Free (Meta, /v1/responses)
• laguna-s-2.1-free — Laguna S 2.1 Free

3. 注册方法

1. 访问 https://opencode.ai/zen → 登录（GitHub/Google）
2. 添加账单（$20 预付，+$1.23 手续费）—— 获取 Zen 网关 Key 必需
3. 从控制台复制 API Key → 作为 Bearer Token 使用

4. 使用方法

Base URL: https://opencode.ai/zen/v1

同上，big-pickle 等 7 个模型走 /v1/chat/completions，muse-spark-1.2-contributor-free 走 /v1/responses。

5. 实测数据

2026-08-30 无鉴权 GET /v1/models 返回 8 个 *-free + big-pickle 模型；go/v1/models 返回 30+ 模型（含付费）。详见 internal/handlers/official-sources.json 的 opencode-zen 条目。

6. 注意事项

• 免费期数据可能被用于模型改进（Big Pickle / MiMo 等）；NVIDIA 免费端点仅记录会话用于安全与改进，不关联身份；Muse Spark Contributor 会用 prompt/completion 训练未来 Meta 模型。
• 免费期结束后自动按 Zen 定价计费（透明定价，无加价，仅收处理费）。

opencode.ai Zen Free API Details

Verified on 2026-08-30 00:00 +08:00 (source: https://opencode.ai/docs/zen + https://opencode.ai/zen/v1/models no-auth verified) Channel slug:opencode-zen | Detail URL:/free-llm-api-rankings/opencode-zen | Base URL:https://opencode.ai/zen/v1

1. Channel Overview

opencode.ai Zen is a curated aggregator for coding agents by the OpenCode team. It bundles 8 anonymous/free models (Big Pickle stealth, MiMo-V2.5, Hy3, Ling 3.0 Fin, Nemotron 3 Ultra/Lightning, Muse Spark Contributor, Laguna) at zero cost for a limited feedback-collection period. All models are exposed via OpenAI-compatible /v1/chat/completions (and /v1/responses for Muse Spark). Requires an opencode.ai Zen gateway key (free, $20 pay-as-you-go with auto top-up). 60 RPM, 200K context.

2. Verified Supported Models

Scope: models_total = 8 (opencode.ai /v1/models no-auth verified + docs cross-verified); chat = 8;all anonymous/free limited-time models

Anonymous/Free Models (8)

• big-pickle — Big Pickle stealth anonymous
• mimo-v2.5-free — MiMo-V2.5 Free
• hy3-free — Hy3 Free
• ling-3.0-flash-fin-free — Ling 3.0 Flash Fin Free
• nemotron-3-ultra-free — Nemotron 3 Ultra Free (NVIDIA free endpoint)
• nemotron-3.5-lightning-free — Nemotron 3.5 Lightning Free (NVIDIA free endpoint)
• muse-spark-1.2-contributor-free — Muse Spark 1.2 Contributor Free (Meta, /v1/responses)
• laguna-s-2.1-free — Laguna S 2.1 Free

3. Registration

1. Visit https://opencode.ai/zen → Sign in (GitHub/Google)
2. Add billing ($20 pay-as-you-go, +$1.23 processing fee) — required to obtain Zen gateway key
3. Copy API key from dashboard → use as Bearer token

4. Usage

Base URL: https://opencode.ai/zen/v1

# List models (no auth required for discovery)
curl https://opencode.ai/zen/v1/models

# Chat completion (requires Zen key)
curl https://opencode.ai/zen/v1/chat/completions \
  -H "Authorization: Bearer $ZEN_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"big-pickle","messages":[{"role":"user","content":"hello"}]}'

# Muse Spark via Responses API
curl https://opencode.ai/zen/v1/responses \
  -H "Authorization: Bearer $ZEN_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"muse-spark-1.2-contributor-free","input":"hello"}'

5. Verified Metrics

2026-08-30 no-auth GET /v1/models returns 8 *-free + big-pickle models; go/v1/models returns 30+ models (incl. paid). See internal/handlers/official-sources.json opencode-zen entry.

6. Notes

• Free-period data may be used for model improvement (Big Pickle / MiMo etc.); NVIDIA free endpoints only log sessions for safety/improvement without identity linkage; Muse Spark Contributor uses prompts/completions to train future Meta models.
• After the free period, billing follows Zen transparent pricing (no markup, processing fee only).