FlexAI Token Factory Pricing
Per-token pricing for 31 open models on FlexAI's Token Factory — an OpenAI-compatible serverless inference API. Every chat model also has a discounted cached-input rate, typically 85% below the standard input price.
Last updated: Nov 21, 2026
Get our weekly newsletter on pricing changes, new releases, and tools.
FlexAI free tier & credits
No ongoing free tier — see credits & cheapest models — see the full FlexAI free tier guide
Provider | Model | Input/1M | Output/1M | Cached input/1M | Context |
|---|---|---|---|---|---|
$0.018 | $0.090 | $0.00 | 131072 | ||
$0.023 | $0.030 | $0.00 | 131072 | ||
$0.030 | $0.150 | $0.00 | 131072 | ||
$0.050 | $0.080 | $0.01 | 131072 | ||
QW | Qwen3 Reranker 8B | $0.050 | $0.000 | - | 40960 |
NV | $0.060 | $0.160 | $0.01 | 1048576 | |
$0.067 | $0.290 | $0.01 | 262144 | ||
QW | $0.070 | $0.280 | $0.01 | 262144 | |
DS | deepseek/deepseek-v4.1-flash-1017 | $0.080 | $0.605 | $0.01 | 1048576 |
Z | $0.088 | $0.250 | $0.01 | 1048576 | |
QW | $0.100 | $0.150 | $0.01 | 262144 | |
QW | $0.100 | $0.950 | $0.01 | 262144 | |
DS | deepseek-ai/DeepSeek-V4-Flash-0731 | $0.130 | $0.260 | $0.02 | 1048576 |
$0.140 | $0.400 | $0.02 | 262144 | ||
Z | $0.140 | $0.860 | $0.02 | 131072 | |
$0.140 | $0.280 | $0.00 | 1048576 | ||
$0.140 | $0.800 | - | 131072 | ||
MM | $0.150 | $0.600 | $0.01 | 1048576 | |
QW | $0.150 | $0.470 | $0.02 | 262144 | |
Llama Guard 4 12B | $0.180 | $0.180 | $0.03 | 16384 | |
QW | $0.200 | $2.400 | $0.03 | 262144 | |
QW | $0.200 | $2.500 | $0.03 | 262144 | |
Llama 3.3 70B Instruct | $0.220 | $0.500 | $0.03 | 131072 | |
QW | $0.250 | $2.000 | $0.03 | 262144 | |
$0.350 | $1.500 | $0.04 | 131072 | ||
Z | $0.600 | $2.000 | $0.09 | 1048576 |
Pricing from FlexAI. Token prices per 1M tokens.
About FlexAI Token Factory
Token Factory is FlexAI's managed inference API for open models — "one key, every model, billed per token." It's OpenAI-compatible, so pointing an existing client at it is usually just a base-URL change, and it spans chat, coding, reasoning, vision, embedding, image, transcription and speech models under a single account.
What stands out against other serverless providers is the published cached-input rate on every chat model — typically about 15% of the normal input price. For workloads with a large repeated prefix (long system prompts, RAG context, multi-turn agents) that's where most of the real saving comes from, and it's listed per model in the table above rather than buried in a footnote.
FlexAI Pricing FAQ
How much does FlexAI Token Factory cost?
Token Factory bills per token with no subscription or token packs. Input prices start around $0.018 per million tokens for smaller open models. See the live table above for current per-model rates.
What is the cached input price?
FlexAI publishes a separate, much cheaper rate for cached input tokens — typically about 15% of the standard input price. If your prompts share a long common prefix, most of your input cost can fall to that rate.
Is Token Factory OpenAI-compatible?
Yes. It exposes an OpenAI-compatible API, so most SDKs and tools work by changing the base URL and API key.
What model types does FlexAI support?
Chat, coding, reasoning and vision models priced per token, plus embeddings, image generation (per image), transcription (per minute) and text-to-speech (per million characters). Use the category tabs above to switch between them.
