Skip to content
PricingSaaS
Sign in
Back
Pricing briefing
TTogether
Together
PACKAGING CHANGE

Together added Kimi K3.1 Pro and Ripple-65B Preview to its model catalog.

By , Co-founder, PricingSaaS

Published

Together added two new models to its serverless inference catalog.

See before and after

Together’s pricing page · Change detected

What changed

Together added two new models to its serverless inference catalog. Kimi K3.1 Pro lists at $2.00 per 1M input tokens and $10.00 per 1M output tokens, with cached input at $0.20 per 1M tokens, per the pricing page Pulse saved on Nov 5, 2026. Ripple-65B Preview lists at $0.20 per 1M input tokens and $0.60 per 1M output tokens, with cached input at $0.02 per 1M tokens, per the same pricing page Pulse saved on Nov 5, 2026.

Why it matters

For a team evaluating inference providers, a broad and frequently refreshed catalog is itself the product: Together is competing on breadth and update speed, not on a single flagship price. Kimi K3.1 Pro lands as a premium-tier model well above the catalog's cheapest options, while the update's second new model adds a cheaper small-model alternative, giving teams more price points to benchmark against whatever they currently run without switching providers.

Watch whether Together keeps adding premium-tier models at this pace: if the cadence holds, choosing a model, not choosing a provider, becomes the real pricing decision for teams building on inference APIs.

In context

Pulse tracked 40 real pricing changes from Together since Dec 29, 2025, averaging one every 8 days, with NVIDIA B200 and Reflection Beam both added earlier in October. Adding models is Together's single most common move: Pulse has logged 13 features added events among the 58 recorded for the company since Dec 29, 2025, more often than the two cached-token price changes made on Kimi K3 in October (Oct 6, Oct 14). Pulse tracked 325 credits-and-usage pricing changes across 224 companies in the last 180 days, the busiest pattern across the companies it tracks, so two new metered-token models fits a market still adding usage-priced entries rather than trimming them. Together is competing on catalog breadth and update speed more than on any single model's price, carrying a premium-tier addition like Kimi K3.1 Pro alongside a cheaper small-model counterpart rather than racing only to the bottom.

Common questions

What should we check before switching workloads to Kimi K3.1 Pro?

Compare it against the models already in your stack on cost per completed task, not just the per-token rate — a pricier model can still win out if it needs fewer retries or shorter prompts to hit the same quality bar.

Is this a one-off or part of a pattern?

Part of a pattern — see In context for how often Together refreshes its model catalog and how that compares with its price-change cadence.

How does Kimi K3.1 Pro differ from the Kimi K3 already in Together's catalog?

Kimi K3's standard rate is $3.00 in / $15.00 out per 1M tokens with $0.30 cached, per Together's pricing page on Oct 14, 2026 — noticeably pricier than the new Pro variant, which Together is positioning as the cheaper option in the Kimi line rather than a straight upgrade.

Sources