# OpenCrabs Usage Pricing Table — usage_pricing.toml.example
#
# This file is copied to ~/.opencrabs/usage_pricing.toml on first run.
# Edit that file live — changes take effect immediately on next /usage open.
# No restart needed.
#
# Rules:
# - `prefix` is matched as a case-insensitive substring of the model name
# - First match within each provider wins — put specific prefixes before general ones
# - Costs are per 1 million tokens (USD)
# - Add any provider/model — if it's not here, cost shows as $0.00
[providers.anthropic]
entries = [
{ prefix = "claude-fable", input_per_m = 10.0, output_per_m = 50.0 },
{ prefix = "claude-opus-4", input_per_m = 5.0, output_per_m = 25.0 },
{ prefix = "claude-3-opus", input_per_m = 15.0, output_per_m = 75.0 },
{ prefix = "claude-sonnet-4", input_per_m = 3.0, output_per_m = 15.0 },
{ prefix = "claude-3-7-sonnet", input_per_m = 3.0, output_per_m = 15.0 },
{ prefix = "claude-3-5-sonnet", input_per_m = 3.0, output_per_m = 15.0 },
{ prefix = "claude-3-sonnet", input_per_m = 3.0, output_per_m = 15.0 },
{ prefix = "claude-haiku-4", input_per_m = 1.0, output_per_m = 5.0 },
{ prefix = "claude-3-5-haiku", input_per_m = 0.80, output_per_m = 4.0 },
{ prefix = "claude-3-haiku", input_per_m = 0.25, output_per_m = 1.25 },
]
[providers.openai]
entries = [
{ prefix = "gpt-5-nano", input_per_m = 0.10, output_per_m = 0.40 },
{ prefix = "gpt-5-mini", input_per_m = 0.30, output_per_m = 1.20 },
{ prefix = "gpt-5", input_per_m = 1.25, output_per_m = 10.0 },
{ prefix = "gpt-4-turbo", input_per_m = 10.0, output_per_m = 30.0 },
{ prefix = "gpt-4", input_per_m = 30.0, output_per_m = 60.0 },
{ prefix = "o3-mini", input_per_m = 1.10, output_per_m = 4.40 },
{ prefix = "o3", input_per_m = 10.0, output_per_m = 40.0 },
{ prefix = "o1-mini", input_per_m = 1.10, output_per_m = 4.40 },
{ prefix = "o1", input_per_m = 15.0, output_per_m = 60.0 },
]
[providers.minimax]
# MiniMax-M3 Standard tier, Context ≤ 512K — public list price.
# Input $0.6/M, Output $2.4/M, Cache read $0.12/M. The launch 7-day
# 50% off discount is NOT applied here (the provider charges the
# discount automatically during the window; this table tracks list
# price). The 512K–1M context band is "contact sales" (limited
# availability), not wired until it goes public.
entries = [
{ prefix = "minimax-m3", input_per_m = 0.60, output_per_m = 2.40, cache_read_per_m = 0.12 },
{ prefix = "minimax-m2.7-high", input_per_m = 0.60, output_per_m = 2.40 },
{ prefix = "minimax-m2.7", input_per_m = 0.30, output_per_m = 1.20 },
{ prefix = "minimax-m2.5-high", input_per_m = 0.60, output_per_m = 2.40 },
{ prefix = "minimax-m2.5", input_per_m = 0.30, output_per_m = 1.20 },
{ prefix = "minimax-m2.1", input_per_m = 0.30, output_per_m = 1.20 },
{ prefix = "minimax-text-01", input_per_m = 0.20, output_per_m = 1.10 },
{ prefix = "minimax", input_per_m = 0.30, output_per_m = 1.20 },
]
[providers.google]
entries = [
{ prefix = "gemini-2.0-flash", input_per_m = 0.10, output_per_m = 0.40 },
{ prefix = "gemini-1.5-pro", input_per_m = 1.25, output_per_m = 5.0 },
{ prefix = "gemini-1.5-flash", input_per_m = 0.075, output_per_m = 0.30 },
]
[providers.deepseek]
entries = [
{ prefix = "deepseek-r1", input_per_m = 0.55, output_per_m = 2.19 },
{ prefix = "deepseek-v3", input_per_m = 0.27, output_per_m = 1.10 },
{ prefix = "deepseek", input_per_m = 0.27, output_per_m = 1.10 },
]
[providers.meta]
entries = [
{ prefix = "llama-3.3-70b", input_per_m = 0.59, output_per_m = 0.79 },
{ prefix = "llama-3.1-405b", input_per_m = 2.70, output_per_m = 2.70 },
{ prefix = "llama-3.1-70b", input_per_m = 0.52, output_per_m = 0.75 },
{ prefix = "llama-3.1-8b", input_per_m = 0.07, output_per_m = 0.07 },
]
[providers.xai]
entries = [
{ prefix = "grok-3-mini", input_per_m = 0.30, output_per_m = 0.50 },
{ prefix = "grok-3", input_per_m = 3.0, output_per_m = 15.0 },
{ prefix = "grok-2", input_per_m = 2.0, output_per_m = 10.0 },
{ prefix = "grok", input_per_m = 5.0, output_per_m = 15.0 },
]
[providers.moonshot]
entries = [
# Kimi K2.6 — Moonshot AI (≤262k context)
# Cache hit: $0.16/1M, Cache miss: $0.95/1M
{ prefix = "kimi-k2.6", input_per_m = 0.95, output_per_m = 4.0, cache_write_per_m = 0.95, cache_read_per_m = 0.16 },
{ prefix = "kimi-k2-6", input_per_m = 0.95, output_per_m = 4.0, cache_write_per_m = 0.95, cache_read_per_m = 0.16 },
{ prefix = "kimik2.6", input_per_m = 0.95, output_per_m = 4.0, cache_write_per_m = 0.95, cache_read_per_m = 0.16 },
{ prefix = "kimi-k2.5", input_per_m = 0.60, output_per_m = 3.0 },
{ prefix = "kimi-k2-turbo", input_per_m = 1.15, output_per_m = 8.0 },
{ prefix = "kimi-k2", input_per_m = 0.60, output_per_m = 2.50 },
{ prefix = "kimi", input_per_m = 0.60, output_per_m = 2.50 },
]
[providers.mistral]
entries = [
{ prefix = "mistral-large", input_per_m = 2.0, output_per_m = 6.0 },
{ prefix = "mistral-small", input_per_m = 0.10, output_per_m = 0.30 },
{ prefix = "codestral", input_per_m = 0.20, output_per_m = 0.60 },
{ prefix = "mistral", input_per_m = 0.25, output_per_m = 0.25 },
]
[providers.qwen]
entries = [
# Qwen 3.6 Max Preview — DashScope tiered pricing (≤128k: $1.3/$7.8, 128k-256k: $2/$12)
# All variants map here: qwen3.6-max-preview, qwen-3.6-max-preview, qwen-3-6-max-preview, etc.
{ prefix = "qwen-3.6-max-preview", input_per_m = 1.30, output_per_m = 7.80, cache_write_per_m = 1.625, cache_read_per_m = 0.13 },
{ prefix = "qwen3.6-max-preview", input_per_m = 1.30, output_per_m = 7.80, cache_write_per_m = 1.625, cache_read_per_m = 0.13 },
{ prefix = "qwen-3-6-max-preview", input_per_m = 1.30, output_per_m = 7.80, cache_write_per_m = 1.625, cache_read_per_m = 0.13 },
{ prefix = "qwen3-6-max-preview", input_per_m = 1.30, output_per_m = 7.80, cache_write_per_m = 1.625, cache_read_per_m = 0.13 },
{ prefix = "qwen-max-preview", input_per_m = 1.30, output_per_m = 7.80, cache_write_per_m = 1.625, cache_read_per_m = 0.13 },
# Qwen 3.6 Plus (thinking) — DashScope tiered pricing (input ≤256k)
# Output corrected to $3.0 per DashScope official pricing (was $1.50)
{ prefix = "qwen-3.6-plus", input_per_m = 0.50, output_per_m = 3.00, cache_write_per_m = 0.625, cache_read_per_m = 0.05 },
{ prefix = "qwen3.6-plus", input_per_m = 0.50, output_per_m = 3.00, cache_write_per_m = 0.625, cache_read_per_m = 0.05 },
{ prefix = "qwen-plus", input_per_m = 0.50, output_per_m = 3.00, cache_write_per_m = 0.625, cache_read_per_m = 0.05 },
]
[providers.zhipu]
entries = [
# GLM-5.1 — Zhipu AI (≤200k context)
{ prefix = "glm-5.1", input_per_m = 0.95, output_per_m = 3.15 },
# GLM-5 Turbo — Zhipu AI
{ prefix = "glm-5-turbo", input_per_m = 0.95, output_per_m = 3.15 },
{ prefix = "glm-5", input_per_m = 0.95, output_per_m = 3.15 },
{ prefix = "zhipu", input_per_m = 0.95, output_per_m = 3.15 },
]
[providers.opencode]
entries = [
{ prefix = "mimo-v2-pro", input_per_m = 1.0, output_per_m = 3.0 },
{ prefix = "mimo-v2-omni", input_per_m = 0.40, output_per_m = 2.0 },
{ prefix = "nemotron-3-super", input_per_m = 0.10, output_per_m = 0.50 },
{ prefix = "big-pickle", input_per_m = 0.0, output_per_m = 0.0 },
{ prefix = "opencode-zen", input_per_m = 0.0, output_per_m = 0.0 },
{ prefix = "opencode-go", input_per_m = 0.0, output_per_m = 0.0 },
]