Capability

Reasoning models

Models that spend tokens thinking before they answer. They win on multi-step problems — planning, maths, debugging, anything where the first plausible answer is usually wrong — and they cost more per call for exactly that reason.

Models
40
Vendors
3
Largest context
1.0M
With cached input
33/40

They take different parameters

This is the practical trap. Reasoning models reject some of the parameters ordinary chat models expect, and a request that works against one will be refused by the other. odnoga reconciles the request to the model it actually routes to, so switching a route from a chat model to a reasoning model does not mean rewriting the call — which is the whole point of choosing the model by rule instead of in code.

Pay for thinking only where it pays

Thinking tokens are billed like any other output, so a reasoning model on a simple classification task is money set on fire. The pattern that works: a cheap model on the hot path, a reasoning model behind an escalation rule for the small share of requests that need it. Route by rule, measure both with the same evaluation cases, and let the numbers decide.

Your plan

Vendor cost is the list price. Your price is that plus your plan margin — the same figures that appear on your invoice.

Every model here, cheapest first

ModelVendorContextIn / 1MOut / 1MCached inYour price in
GPT-5 nano
gpt-5-nano
OpenAI400K$0.05$0.4$0.005$0.054Details
GPT-5.4 nano
gpt-5.4-nano
OpenAI400K$0.2$1.25$0.02$0.215Details
GPT-5.6 Luna
gpt-5.6-luna
OpenAI400K$0.2$1.20$0.02$0.215Details
Gemini 3.1 Flash-Lite
gemini-3.1-flash-lite
Google AI1.0M$0.25$1.50$0.025$0.269Details
GPT-5 mini
gpt-5-mini
OpenAI400K$0.25$2$0.025$0.269Details
Gemini 3.5 Flash-Lite
gemini-3.5-flash-lite
Google AI1.0M$0.3$2.50$0.03$0.323Details
Grok 3 Mini
grok-3-mini
xAI131K$0.3$0.5$0.075$0.323Details
Gemini 3 Flash Preview
gemini-3-flash-preview
Google AI1.0M$0.5$3$0.05$0.538Details
Gemini 3.1 Flash Image (Nano Banana 2)
gemini-3.1-flash-image
Google AI66K$0.5$3$0.538Details
Gemini 3.6 Flash
gemini-3.6-flash
Google AI1.0M$0.75$3.75$0.075$0.806Details
Gemini 3.7 Flash
gemini-3.7-flash
Google AI1.0M$0.75$3.75$0.075$0.806Details
GPT-5.4 mini
gpt-5.4-mini
OpenAI400K$0.75$4.50$0.075$0.806Details
Gemini Robotics ER 2
gemini-robotics-er-2-preview
Google AI131K$1$5$0.1$1.08Details
grok-build-0.1
grok-build-0.1
xAI256K$1$2$0.2$1.08Details
o3-mini
o3-mini
OpenAI200K$1.10$4.40$0.55$1.18Details
o4-mini
o4-mini
OpenAI200K$1.10$4.40$0.275$1.18Details
Gemini 2.5 Pro
gemini-2.5-pro
Google AI1.0M$1.25$10$0.125$1.34Details
GPT-5
gpt-5
OpenAI400K$1.25$10$0.125$1.34Details
gpt-5.1
gpt-5.1
OpenAI400K$1.25$10$0.125$1.34Details
grok-4.20-0309-reasoning
grok-4.20-0309-reasoning
xAI1M$1.25$2.50$0.2$1.34Details
grok-4.20-multi-agent-0309
grok-4.20-multi-agent-0309
xAI1M$1.25$2.50$0.2$1.34Details
grok-4.3
grok-4.3
xAI1M$1.25$2.50$0.2$1.34Details
Gemini 3.5 Flash
gemini-3.5-flash
Google AI1.0M$1.50$9$0.15$1.61Details
Gemini Omni Flash Preview
gemini-omni-flash-preview
Google AI131K$1.50$9$1.61Details
gpt-5.2
gpt-5.2
OpenAI400K$1.75$14$0.175$1.88Details
GPT-5.3 Codex
gpt-5.3-codex
OpenAI400K$1.75$14$0.175$1.88Details
Gemini 3 Pro Image (Nano Banana Pro)
gemini-3-pro-image
Google AI131K$2$12$2.15Details
Gemini 3.1 Pro Preview
gemini-3.1-pro-preview
Google AI1.0M$2$12$0.2$2.15Details
GPT-5.6 Terra
gpt-5.6-terra
OpenAI400K$2$12$0.2$2.15Details
grok-4.5
grok-4.5
xAI500K$2$6$0.3$2.15Details
grok-4.6
grok-4.6
xAI500K$2$6$0.5$2.15Details
o3
o3
OpenAI200K$2$8$0.5$2.15Details
GPT-5.4
gpt-5.4
OpenAI400K$2.50$15$0.25$2.69Details
Grok 4
grok-4
xAI256K$3$15$0.75$3.23Details
GPT-5.6 Sol
gpt-5.6-sol
OpenAI400K$4$20$0.4$4.30Details
GPT-5.5
gpt-5.5
OpenAI400K$5$30$0.5$5.38Details
gpt-5-pro
gpt-5-pro
OpenAI400K$15$120$16.13Details
gpt-5.2-pro
gpt-5.2-pro
OpenAI400K$21$168$22.58Details
GPT-5.4 Pro
gpt-5.4-pro
OpenAI400K$30$180$32.26Details
GPT-5.5 Pro
gpt-5.5-pro
OpenAI400K$30$180$32.26Details

Your price column is the vendor cost +7%.

One key, every model on this page.

Change the model with a routing rule instead of a deploy, and bill every call to the customer who made it.