Model catalog
48+ models, one endpoint
Every model below is live in RemKey's own registry right now — not a
roadmap, not a partner's catalog we're describing secondhand. The three tiers are what the
classifier auto-routes between; below that, models are split honestly by who actually serves
the request.
Cheap tier (auto-routed) 1
| Model | Provider | $/1M in | $/1M out |
| RK claude-haiku |
RemKey (governed) |
$0.8 |
$4 |
Standard tier (auto-routed) 1
| Model | Provider | $/1M in | $/1M out |
| RK claude-sonnet |
RemKey (governed) |
$3 |
$15 |
Premium tier (auto-routed) 1
| Model | Provider | $/1M in | $/1M out |
| RK claude-opus |
RemKey (governed) |
$15 |
$75 |
Governed catalog 32
Served directly by RemKey on our own infrastructure — margin-priced, single-region, fully
audited, no separate account or key required from you. Requires an explicit
vendor/model passthrough request and the pooled lane enabled on your tenant.
| Model | Provider | $/1M in | $/1M out |
| RK ai21/jamba-1.5-large |
RemKey (governed) |
$0.5 |
$1.5 |
| RK ai21/jamba-1.5-mini |
RemKey (governed) |
$0.2 |
$0.6 |
| RK amazon/nova-lite |
RemKey (governed) |
$0.075 |
$0.3 |
| RK amazon/nova-micro |
RemKey (governed) |
$0.035 |
$0.14 |
| RK amazon/nova-pro |
RemKey (governed) |
$0.8 |
$3.2 |
| RK deepseek/deepseek-v3.2 |
RemKey (governed) |
$0.62 |
$1.85 |
| RK google/gemma-3-12b |
RemKey (governed) |
$0.09 |
$0.29 |
| RK google/gemma-3-27b |
RemKey (governed) |
$0.23 |
$0.38 |
| RK google/gemma-3-4b |
RemKey (governed) |
$0.04 |
$0.08 |
| RK meta/llama3-70b |
RemKey (governed) |
$0.8 |
$2.4 |
| RK meta/llama3-8b |
RemKey (governed) |
$0.3 |
$0.6 |
| RK minimax/minimax-m2 |
RemKey (governed) |
$0.3 |
$1.2 |
| RK minimax/minimax-m2.1 |
RemKey (governed) |
$0.3 |
$1.2 |
| RK minimax/minimax-m2.5 |
RemKey (governed) |
$0.3 |
$1.2 |
| RK mistral/devstral-2-123b |
RemKey (governed) |
$0.4 |
$2 |
| RK mistral/magistral-small |
RemKey (governed) |
$0.5 |
$1.5 |
| RK mistral/ministral-3-14b |
RemKey (governed) |
$0.2 |
$0.2 |
| RK mistral/ministral-3-3b |
RemKey (governed) |
$0.1 |
$0.1 |
| RK mistral/ministral-3-8b |
RemKey (governed) |
$0.15 |
$0.15 |
| RK mistral/mistral-large-3 |
RemKey (governed) |
$0.5 |
$1.5 |
| RK moonshot/kimi-k2-thinking |
RemKey (governed) |
$0.6 |
$2.5 |
| RK moonshot/kimi-k2.5 |
RemKey (governed) |
$0.6 |
$3 |
| RK nvidia/nemotron-nano-30b |
RemKey (governed) |
$0.06 |
$0.24 |
| RK nvidia/nemotron-super-120b |
RemKey (governed) |
$0.15 |
$0.65 |
| RK openai/gpt-oss-safeguard-120b |
RemKey (governed) |
$0.15 |
$0.6 |
| RK openai/gpt-oss-safeguard-20b |
RemKey (governed) |
$0.07 |
$0.2 |
| RK qwen/qwen3-coder-next |
RemKey (governed) |
$0.5 |
$1.2 |
| RK qwen/qwen3-next-80b |
RemKey (governed) |
$0.15 |
$1.2 |
| RK qwen/qwen3-vl-235b |
RemKey (governed) |
$0.53 |
$2.66 |
| RK zai/glm-4.7 |
RemKey (governed) |
$0.6 |
$2.2 |
| RK zai/glm-4.7-flash |
RemKey (governed) |
$0.07 |
$0.4 |
| RK zai/glm-5 |
RemKey (governed) |
$1 |
$3.2 |
OpenRouter passthrough catalog 13
Served through your own OpenRouter key when a passthrough model isn't (yet) in our
governed catalog above — zero margin to us, your key, your bill.
| Model | Provider | $/1M in | $/1M out |
| OR deepseek/deepseek-chat |
openrouter (your key) |
$0.27 |
$1.1 |
| OR deepseek/deepseek-r1 |
openrouter (your key) |
$0.55 |
$2.19 |
| OR google/gemini-2.5-flash |
openrouter (your key) |
$0.3 |
$2.5 |
| OR google/gemini-2.5-pro |
openrouter (your key) |
$1.25 |
$10 |
| OR meta-llama/llama-3.1-405b-instruct |
openrouter (your key) |
$0.8 |
$0.8 |
| OR meta-llama/llama-3.3-70b-instruct |
openrouter (your key) |
$0.13 |
$0.4 |
| OR mistralai/mistral-large |
openrouter (your key) |
$2 |
$6 |
| OR openai/gpt-4.1 |
openrouter (your key) |
$2 |
$8 |
| OR openai/gpt-4.1-mini |
openrouter (your key) |
$0.4 |
$1.6 |
| OR openai/gpt-4o |
openrouter (your key) |
$2.5 |
$10 |
| OR openai/gpt-4o-mini |
openrouter (your key) |
$0.15 |
$0.6 |
| OR openai/o3-mini |
openrouter (your key) |
$1.1 |
$4.4 |
| OR qwen/qwen-2.5-72b-instruct |
openrouter (your key) |
$0.35 |
$0.4 |
Pricing shown is $ per 1M tokens, snapshot at generation time —
this page always reflects the live registry, so it can't drift from what a request would
actually be billed.