Together adds public HGX B300 GPU pricing: $4.99–$9.99/GPU/hr
Together AI made NVIDIA HGX B300 GPU pricing public on its GPU Clusters table: $4.99/GPU/hr preemptible and $9.99/GPU/hr on-demand pay-as-you-go.
Together’s pricing page · Change Detected
WHY IT MATTERS
A public B300 rate lets buyers comparison-shop Together against other GPU cloud providers without a sales call, which typically signals rising competitive pressure in that hardware tier. Watch whether Together's next move on B300 is a price cut, following the same trajectory H200 pricing took after going public.
BY THE NUMBERS
- 4th time Together has revealed previously-gated pricing
- 37 changes in the last 365 days, ~7 days apart on average
- Previous change: Qwen3.7-Max and Qwen3.8 Flash prices cut 40%, 2 days earlier
- Price transparency moves appear in 208 of 5,664 tracked changes (150 companies) in 180 days
THEIR RATIONALE
Together frames this as transparency on an existing product line rather than a new offering — the B300 row already existed, just without a visible number. It comes two days after Together's last change, consistent with a company that reprices or restructures something on its platform every 7 days on average, the fastest cadence of any company in this batch.
THE SIGNAL
This is Together's fourth time revealing previously-gated pricing, following the same move for H200 GPUs in May and cached-token pricing in June. The pattern says something about Together's go-to-market: newer, higher-end hardware launches behind a 'contact us' wall, then gets a public number once demand and margins are established enough to compete openly. Price transparency itself is a fast-growing category across the index — 208 changes, 150 companies in six months — as vendors from Shopify to Claris drop opaque pricing for public rate cards.
IN CONTEXT
Together has made hidden pricing public three times before this one — most notably revealing H200 GPU on-demand rates in May, which went from 'Contact us' to $6.79/hour. The B300 disclosure follows the same script: a premium GPU tier launches gated, then gets a public number once Together is ready to compete on price. It also comes just two days after Together cut Qwen3.7-Max and Qwen3.8 Flash inference prices 40%, part of a volatile few weeks that saw the same model hiked 60% in September before this reversal. Price transparency is picking up across the index too, with Shopify disclosing Plus payment processing rates and Claris publishing FileMaker Cloud prices in the same week.