Quan 3.4 is now running on a stronger and faster reasoning engine. For developers building stock analysis tools, finance copilots, research workflows, and market-data products, this update delivers the combination we have been working toward: higher quality, faster responses, and a lower price for every API user.
We did not make this change because a new model number appeared. We made it because our concrete testing showed that the updated runtime was materially better for the tasks Quan is built to handle: multi-step financial reasoning, grounded market context, comparisons, valuation questions, and responses that need to keep assumptions and calculations aligned.
Quan should improve as an API product without making developers choose between better answers and a more manageable bill.
What changed in Quan 3.4?
Quan 3.4 remains the same finance-focused API surface and model name. The underlying reasoning runtime has been upgraded, so existing integrations can keep calling quan-3.4 while receiving the improved model behavior.
This is important for teams that have already built around the Quan 3.4 API. You do not need to redesign your request format, change your API key, or migrate to a new product name to benefit from the upgrade. The improvement arrives behind the existing Quan 3.4 interface.
Higher-quality financial AI reasoning
Finance questions are rarely one-step questions. A useful answer may need to identify the right company context, separate current facts from interpretation, compare several periods, reconcile a calculation, and explain uncertainty without losing the original question.
In our Quan testing, the updated runtime handled these multi-step tasks more reliably than the previous version. The practical result is stronger performance for:
- Stock analysis API workflows: company comparisons, valuation context, catalysts, risks, and market-move explanations.
- Financial calculations: assumption-aware forecasts, margin analysis, portfolio math, and scenario comparisons.
- Grounded research: answers that use current evidence and distinguish verified facts from model interpretation.
- Developer products: customer-facing finance copilots, internal research tools, and agent workflows that need consistent output.
Faster responses for real API workloads
Speed matters when Quan sits inside an application. A slower financial AI response can make a research queue feel stuck, interrupt an agent loop, or make a user abandon a market question before the answer arrives.
The upgraded runtime gives Quan a faster foundation for the repeated calls common in financial AI products. That means more responsive stock research, quicker iteration during development, and better flow for applications that combine quote context, calculations, and explanation.
New Quan 3.4 API pricing
Because the new runtime is more cost-efficient, we are passing the savings to all Quan 3.4 API users. The new token rates are:
Quan 3.4 developer API
This is a direct price reduction on the advertised Quan 3.4 rates. For example, 1 million input tokens and 1 million output tokens now cost $17.50 total: $2.50 for input plus $15.00 for output. Input and output are metered separately, so users pay only for the tokens their application actually consumes.
Lower-cost live internet grounding
We found a way to reduce the cost of live internet and search grounding. Paid grounding is now passed through at $0.015 per executed search. Quan 3.4 L still includes up to two searches in its free allowance; additional searches use the prepaid balance.
This separates model-token pricing from live research costs. Developers can estimate their application more clearly: token usage covers the reasoning and response, while an executed live search is a distinct $0.015 line item when grounding is enabled.
What this means for developers
If you already use the Quan 3.4 API, the migration is designed to be low-friction. Keep your existing model value, request structure, and API integration. Your main change is the lower price applied to new usage.
If you are evaluating a financial AI API for a new product, this is a good time to test Quan 3.4 against the actual tasks your users will ask. Try multi-company comparisons, earnings explanations, portfolio risk reviews, valuation questions, SEC filing analysis, and grounded market research—not just short generic prompts.
We believe the best model upgrade is one users can feel in the answer and see in the bill. Quan 3.4 is now better equipped for serious finance workflows, faster to use in production, and cheaper to operate at the API layer.
Start building with the updated Quan 3.4
Create a StockUp developer API key, review the Quan 3.4 API developer guide, or explore the current API pricing. Quan is built for finance-native applications that need grounded stock analysis, quantitative reasoning, and a practical path from prototype to production.