Why Output Speed Matters in a Cost Comparison
Once you've narrowed a model shortlist on the main TokenRate calculator, the Compare Prices side-by-side view is where you stack them for a decision. Each row shows the provider, the model ID (the one you'd paste into your SDK), per-1M input and output costs, the context window, and the blended quality score. Most teams compare LLMs on per-token input price and call it a day. Output Speed is one of the under-attended dimensions that shows up in the Compare Prices grid — and ignoring it is where the surprise bills come from. Output speed (tokens per second) isn't in the Compare Prices grid columns — but it's correlated with model size and tier in ways that matter for routing decisions. Related reading: quality per dollar LLM ranking 2026, LLM color-coded quality badges explained, and why the cheapest LLM isn't always the best value.
How Output Speed Appears in the Grid
In the grid, the Fast tier is the rough proxy for speed: smaller models output more tokens per second on the same hardware. Haiku 4.5, Gemini Flash, GPT-4o mini — all output 100+ tps in practice.
Reading the Spread
Reasoning models (o3, R1, Sonnet thinking) are intentionally slower because chain-of-thought adds compute. The Compare Prices grid groups them in the Reasoning tier so the speed expectation is calibrated.
Practical Implications
For applications where output speed matters (live chat, real-time agents), the right Compare Prices workflow is: filter to Fast tier first, then compare pricing within the tier. Don't pick a Reasoning model for a streaming chatbot.
Workflow: From Grid to Decision
For up-to-date throughput numbers, cross-reference Artificial Analysis's speed leaderboard alongside the Compare Prices grid.