Reviewed · Published by RiverX
Separate the model from the coding agent
The model generates responses and tool decisions; the coding-agent runtime manages the development interaction. RiverX supports model selection and Codex/OpenCode runtimes, subject to configuration. A model's catalog listing does not guarantee it works with every agent or every tool. Confirm the exact combination in the account where you will build.
Match capabilities to a concrete task
For screenshot-led work, check image-input support. For repository edits, verify tool calling. For a large refactor, inspect the context window and how the runtime supplies files. For routine copy changes, evaluate whether a lower-cost option produces an acceptable result. Avoid treating the largest context number or newest model name as a universal recommendation.
Run a controlled comparison
Create a starting commit and one acceptance test, such as adding validation to a request form without changing its successful submission. Start each trial from that same code and provide the same relevant context. Record errors, changed files, test results and usage. If you change models during a task, check the available context and restate constraints when necessary.
Read provider prices in context
Input, output and cached-input prices can be different, and some catalog fields may be missing. Provider prices describe recorded inference rates, not the final charge for an entire feature. Compare actual wallet usage and accepted work. Repeat the evaluation when the provider or model version changes; this page does not promise benchmark results or uninterrupted access.
Model choice across the market
Model and agent selection also appears in competitor products. Bolt currently documents agent choices with underlying routing, while v0, Replit and Base44 document plan-dependent model controls. Use the linked product comparisons to inspect those differences. RiverX's value should be evaluated through its actual catalog, runtime and development workflow rather than an unsupported claim that alternatives offer no choice.
Sources: Bolt agents · v0 model selection · Replit model selector · Base44 AI models
Can I choose an AI model in RiverX?
Yes. RiverX exposes selectable models and a public directory of recorded capabilities and pricing. Account, provider and runtime availability still matter.
Does every catalog model support tools or images?
No such guarantee is made. Check the exact entry's capabilities and test it with the intended coding runtime.
Does switching models guarantee better results?
No. Compare a repeatable task with explicit acceptance criteria, including retries, review time and actual usage cost.
Are RiverX's model-directory prices final build prices?
No. They describe provider rates where recorded. A build can involve multiple requests, and wallet charges and checkout fees follow the billing rules.
Put the pieces together
Models, instructions and starting points for your next project.