Models

Models & live prices

Every model Flow AI routes to, with the live clearing price vs the published rate. Prices float within a floor↔ceiling band as supply and demand move (↓ = getting cheaper). Pass model: "auto" and Flow AI serves the cheapest model that can do the job — or pin any model below.

ModelClearing $/M (in / out)Live range (in $/M)Published $/MSavings
deepseek-v4-flashpinnable$0.13 / $0.27 $0.13 – $0.14$0.14 / $0.285% off
minimax-m2.7pinnable$0.04 / $0.11 $0.03 – $0.25$0.25 / $1.0088.4% off
minimax-m3pinnable$0.05 / $0.14 $0.03 – $0.30$0.30 / $1.2087.3% off
deepseek-v4-propinnable$0.43 / $0.87 $0.43 – $0.43$0.43 / $0.87
qwen3.7-plus$0.45 / $1.80 $0.45 – $0.45$0.45 / $1.80
claude-opus-4.8pinnable$15.00 / $75.00 $15.00 – $15.00$15.00 / $75.00
nemotron-3-ultrapinnable$0.00 / $0.00 $0.00 – $0.00$0.00 / $0.00
kimi-k2pinnable$0.58 / $2.29 $0.58 – $0.58$0.58 / $2.29
qwen3-27bpinnable$0.20 / $0.80 $0.20 – $0.20$0.20 / $0.80
gpt-oss-120bpinnable$0.04 / $0.19 $0.04 – $0.04$0.04 / $0.19
gpt-4o-minipinnable$0.07 / $0.30 $0.07 – $0.15$0.15 / $0.6050% off
gpt-4opinnable$2.50 / $10.00 $2.50 – $2.50$2.50 / $10.00
gemini-2.5-flashpinnable$0.30 / $2.50 $0.30 – $0.30$0.30 / $2.50
claude-sonnet-4.6pinnable$3.00 / $15.00 $3.00 – $3.00$3.00 / $15.00
mimo-v2.5pinnable$0.06 / $0.24 $0.06 – $0.30$0.30 / $1.2080% off
mimo-v2.5-propinnable$0.60 / $2.40 $0.60 – $0.60$0.60 / $2.40
qwen3-8bself-hostpinnable$0.08 / $0.16 $0.08 – $0.08$0.08 / $0.16
qwen3-14bself-host$0.10 / $0.20 $0.10 – $0.10$0.10 / $0.20
qwen3-32bself-host$0.15 / $0.30 $0.15 – $0.15$0.15 / $0.30
mistral-smallself-hostpinnable$0.12 / $0.24 $0.12 – $0.12$0.12 / $0.24
qwen2.5-7bself-hostpinnable$0.08 / $0.16 $0.08 – $0.08$0.08 / $0.16
llama3.3-70bself-hostpinnable$0.25 / $0.50 $0.25 – $0.25$0.25 / $0.50
qwen2.5-coder-7bself-host$0.08 / $0.16 $0.08 – $0.08$0.08 / $0.16
qwen2.5-coder-14bself-host$0.10 / $0.20 $0.10 – $0.10$0.10 / $0.20
qwen2.5-coder-32bself-hostpinnable$0.15 / $0.30 $0.15 – $0.15$0.15 / $0.30
devstralself-host$0.12 / $0.24 $0.12 – $0.12$0.12 / $0.24
gemma3-4bself-hostpinnable$0.05 / $0.10 $0.05 – $0.05$0.05 / $0.10
phi-4-miniself-hostpinnable$0.04 / $0.08 $0.04 – $0.04$0.04 / $0.08
qwen2.5-3bself-host$0.05 / $0.10 $0.05 – $0.05$0.05 / $0.10
llama3.2-3bself-host$0.05 / $0.10 $0.05 – $0.05$0.05 / $0.10
gemma2-2bself-host$0.04 / $0.08 $0.04 – $0.04$0.04 / $0.08
glm-4.6$0.60 / $2.20 $0.60 – $0.60$0.60 / $2.20
glm-4.7$0.60 / $2.20 $0.60 – $0.60$0.60 / $2.20
glm-5.1$1.40 / $4.40 $1.40 – $1.40$1.40 / $4.40
glm-5.2$1.40 / $4.40 $1.40 – $1.40$1.40 / $4.40
kimi-for-coding$0.18 / $0.90 $0.18 – $0.60$0.60 / $3.0070% off
gpt-5.5$4.50 / $27.00 $4.50 – $5.00$5.00 / $30.0010% off
minimax-m2$0.05 / $0.20 $0.05 – $0.25$0.25 / $1.0080% off
minimax-m2.1$0.06 / $0.19 $0.06 – $0.29$0.29 / $0.9580% off
minimax-m2.1-highspeed$0.06 / $0.19 $0.06 – $0.29$0.29 / $0.9580% off
minimax-m2.5$0.03 / $0.23 $0.03 – $0.15$0.15 / $1.1580% off
minimax-m2.5-highspeed$0.04 / $0.34 $0.04 – $0.15$0.15 / $1.1570% off
gpt-5-nano / / 50% off
gpt-5.1-codex-mini / / 55% off
gpt-5.5-mini / / 55% off
gpt-5.3-codex-spark / / 10% off
gpt-4.1-mini / / 50% off
gpt-5.5-codex / / 30% off
gpt-5-codex / / 30% off
codex-mini-latest / / 40% off
o4-mini / / 40% off
gpt-5-mini / / 71% off

Panel Mode — need a specific model?

Smart routing is the default — but judge ensembles, cross-model evals, and reproducible research need exact models. Models tagged pinnable can be pinned strictly (model: "pin:gpt-4o-mini") or queried as a panel — one prompt across up to 10 models in a single POST /v1/panel call, each answer labeled with the model that really served it, never a silent substitution. Enable it per key in the dashboard. Panel Mode docs →

Prices are per million tokens and update live as supply changes. See the docs to get started, or compare Flow AI to OpenRouter.