Skip to content
tokenblackhole.
Menu
Read caveatsSource: mixed

InferX

10x on selected models — but the context stops at 262k

InferX's own catalog lists a promotional rate at a tenth of the crossed-out original on selected models, which is where the 10x comes from. The catch is not the price: most endpoints cap context at 262,000 tokens, against 1M on DeepSeek's own API.

Open InferX

Tiers

  • Promotional rate — the active price is 10% of the crossed-out list rate, per $1 spent$1.00 → $10.0010x

What the headline leaves out

  • The context cap is the real trade. Long-context work that fits in DeepSeek's 1M window will not fit in 262k, so this is not a like-for-like swap for every workload.
  • The 10x applies to models carrying the promotional price; the rest of the 200+ catalog is billed at its listed rate.
  • Promotional pricing can end without notice — the catalog is the only place it is stated.

Source: https://inferx.net/models

Runs its own inference runtime — snapshot-restored cold starts, scale-to-zero, OpenAI-compatible. Hosted endpoints carry a promotional rate on selected models; the catalog shows the original rate crossed out beside the active one.