SonsofAnarchy
Royal
Royal
5
YEARS
5 YEARS OF SERVICE
Manylane.chat
UNLIMITED requests. One key. Three protocols. Flat rate.
No request counts. No token metering. No per-token invoice. You pay for TIME, not tokens — $5 buys 24 hours of as much as you can push through it.
Hey everyone,
I built Manylane because per-token billing and watching a usage meter gets old fast. Manylane flips the model: you pay for time, not tokens. An active membership means unlimited requests — keep building, keep iterating, keep the terminal warm — and there is nothing to run out of.
It's an independent API gateway: one key, a launch catalog of strong models, and whichever protocol your tool already speaks. No SDK rewrites, no proxy shims.
▸ Unlimited, for real — what that means
- No daily request count — there is no allowance to exhaust
- No monthly token quota — nobody is counting your tokens
- No per-request or per-token price — your membership is never converted into a hidden credit balance
- Requests run unlimited for the entire membership period — day, week, or month, playground included
Unlimited means no usage allowance — it does not mean infinite simultaneous throughput. Each account runs 10 concurrent requests (clean 429 + Retry-After beyond that; slots free up as requests finish). That is the only limit — no body caps, no per-request deadlines, no hidden metering. Full details on the site's fair-use page.
▸ One key, three native endpoints
- OpenAI Chat Completions — /v1/chat/completions
- OpenAI Responses — /v1/responses
- Anthropic Messages — /v1/messages
All streaming (SSE), all real, all on the same vl_sk_... key. Crypto checkout (BTC, USDT, TON and more) with server-verified activation, no auto-renewal, no email required — your account code is your credential.
▸ Launch model lineup
- kimi-k3 — Kimi K3 · reasoning, research, multi-step work
- qwen3.8-max-0902 — Qwen 3.8 Max · everyday and complex tasks
- qwen3.8-flash — Qwen 3.8 Flash · rapid iteration
- deepseek-v4-pro-0813 — DeepSeek V4 Pro · code and technical work
- deepseek-v4.1-flash — DeepSeek V4.1 Flash · fast workflows
- glm-5.3 — GLM 5.3 · coding and planning
Cursor users: pick the manylane-<model-id> aliases in the custom model picker.
▸ Drop-in setup
Claude Code (env, then just run claude):
Code:
Codex CLI — merge into your ~/.codex/config.toml:
Code:
Then export MANYLANE_API_KEY in your terminal and run codex. (The agents / enabled = false line matters on current Codex builds — the gateway takes function tools only.)
Any OpenAI / Anthropic SDK, or plain curl:
Code:
▸ Pricing (launch) — unlimited requests on every plan
- Day pass — $5 · 24 hours · unlimited requests
- Weekly — $10 · 7 days · unlimited requests (~$1.43/day)
- Monthly — $30 · 30 days · unlimited requests ($1.00/day)
Flat rate, paid in crypto, activated automatically after confirmation. No auto-renewal — you decide if and when to extend.
▸ Straight talk
- No silent model substitution — if the selected model is unavailable or rejects a parameter, the API returns an error. It never quietly routes you to a cheaper model.
- Context and output limits are per-model — the directory on the site lists them.
- Model status comes from real requests through the gateway, not synthetic pings.
▸ Support
Ticket system is built into the site — no Discord roulette, no DM chase.
▸ Fine print
Manylane is an independent interface and API gateway for the models in its catalog — not affiliated with the owners of those model brands. Model availability is subject to upstream capacity. Model output can be wrong; review it before you rely on it.
Unlimited. Flat-rate. One key.
→ manylane.chat
Questions and feedback welcome right here in the thread — I'll keep an eye on it.
Manylane — UNLIMITED AI API, one key, flat $5/day · Kimi K3 · Qwen · DeepSeek · GLM → manylane.chat
UNLIMITED requests. One key. Three protocols. Flat rate.
No request counts. No token metering. No per-token invoice. You pay for TIME, not tokens — $5 buys 24 hours of as much as you can push through it.
Hey everyone,
I built Manylane because per-token billing and watching a usage meter gets old fast. Manylane flips the model: you pay for time, not tokens. An active membership means unlimited requests — keep building, keep iterating, keep the terminal warm — and there is nothing to run out of.
It's an independent API gateway: one key, a launch catalog of strong models, and whichever protocol your tool already speaks. No SDK rewrites, no proxy shims.
▸ Unlimited, for real — what that means
- No daily request count — there is no allowance to exhaust
- No monthly token quota — nobody is counting your tokens
- No per-request or per-token price — your membership is never converted into a hidden credit balance
- Requests run unlimited for the entire membership period — day, week, or month, playground included
Unlimited means no usage allowance — it does not mean infinite simultaneous throughput. Each account runs 10 concurrent requests (clean 429 + Retry-After beyond that; slots free up as requests finish). That is the only limit — no body caps, no per-request deadlines, no hidden metering. Full details on the site's fair-use page.
▸ One key, three native endpoints
- OpenAI Chat Completions — /v1/chat/completions
- OpenAI Responses — /v1/responses
- Anthropic Messages — /v1/messages
All streaming (SSE), all real, all on the same vl_sk_... key. Crypto checkout (BTC, USDT, TON and more) with server-verified activation, no auto-renewal, no email required — your account code is your credential.
▸ Launch model lineup
- kimi-k3 — Kimi K3 · reasoning, research, multi-step work
- qwen3.8-max-0902 — Qwen 3.8 Max · everyday and complex tasks
- qwen3.8-flash — Qwen 3.8 Flash · rapid iteration
- deepseek-v4-pro-0813 — DeepSeek V4 Pro · code and technical work
- deepseek-v4.1-flash — DeepSeek V4.1 Flash · fast workflows
- glm-5.3 — GLM 5.3 · coding and planning
Cursor users: pick the manylane-<model-id> aliases in the custom model picker.
▸ Drop-in setup
Claude Code (env, then just run claude):
Code:
Code:
export ANTHROPIC_BASE_URL="https://manylane.chat"
export ANTHROPIC_AUTH_TOKEN="$MANYLANE_API_KEY"
export ANTHROPIC_MODEL="kimi-k3"
export ANTHROPIC_DEFAULT_SONNET_MODEL="kimi-k3"
export ANTHROPIC_DEFAULT_OPUS_MODEL="kimi-k3"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="qwen3.8-flash"
export CLAUDE_CODE_SUBAGENT_MODEL="kimi-k3"
Codex CLI — merge into your ~/.codex/config.toml:
Code:
Code:
model = "glm-5.3"
model_provider = "manylane"
web_search = "disabled"
model_supports_reasoning_summaries = false
[agents]
enabled = false
[model_providers.manylane]
name = "Manylane"
base_url = "https://manylane.chat/v1"
env_key = "MANYLANE_API_KEY"
wire_api = "responses"
requires_openai_auth = false
supports_websockets = false
Then export MANYLANE_API_KEY in your terminal and run codex. (The agents / enabled = false line matters on current Codex builds — the gateway takes function tools only.)
Any OpenAI / Anthropic SDK, or plain curl:
Code:
Code:
curl https://manylane.chat/v1/chat/completions \
-H "Authorization: Bearer $MANYLANE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"kimi-k3","messages":[{"role":"user","content":"hello"}]}'
curl https://manylane.chat/v1/messages \
-H "x-api-key: $MANYLANE_API_KEY" \
-H "anthropic-version: 2023-06-01" \
-H "Content-Type: application/json" \
-d '{"model":"kimi-k3","max_tokens":256,"messages":[{"role":"user","content":"hello"}]}'
▸ Pricing (launch) — unlimited requests on every plan
- Day pass — $5 · 24 hours · unlimited requests
- Weekly — $10 · 7 days · unlimited requests (~$1.43/day)
- Monthly — $30 · 30 days · unlimited requests ($1.00/day)
Flat rate, paid in crypto, activated automatically after confirmation. No auto-renewal — you decide if and when to extend.
▸ Straight talk
- No silent model substitution — if the selected model is unavailable or rejects a parameter, the API returns an error. It never quietly routes you to a cheaper model.
- Context and output limits are per-model — the directory on the site lists them.
- Model status comes from real requests through the gateway, not synthetic pings.
▸ Support
Ticket system is built into the site — no Discord roulette, no DM chase.
▸ Fine print
Manylane is an independent interface and API gateway for the models in its catalog — not affiliated with the owners of those model brands. Model availability is subject to upstream capacity. Model output can be wrong; review it before you rely on it.
Unlimited. Flat-rate. One key.
→ manylane.chat
Questions and feedback welcome right here in the thread — I'll keep an eye on it.
Manylane — UNLIMITED AI API, one key, flat $5/day · Kimi K3 · Qwen · DeepSeek · GLM → manylane.chat





