Skip to main content
Available to users on the new pricing plans (currently not available on legacy plans)

​
Overview

Zencoder now has a model selector in the chat input. Use the dropdown to pick which LLM runs your messages. The options you see depend on your plan and entitlements.
  • Currently, the model selector is available to users on the new pricing plans ( not available on legacy plans)
  • Model choices may differ by plan; higher plans unlock additional models
  • Select the model that fits your task from the selector (Auto, Auto+, Haiku 4.5 Parallel Thinking, Sonnet 4, Sonnet 4 Parallel Thinking, Sonnet 4.5 Parallel Thinking, Opus 4.1, Opus 4.1 Parallel Thinking, Gemini 2.5 Pro, GPT-5, GPT-5 Codex, Grok Code Fast 1).
As a truly vendor-agnostic platform, Zencoder supports flagship models from OpenAI, Anthropic, Google, and xAI, allowing you to choose your preferred model while benefiting from our infrastructure, capabilities, and excellent user experience.
Model selection is fixed per chat session. To use a different model, start a new chat.

​
Available models (subject to change)

The Auto model routes across a tuned mix of self-hosted and vendor models for the best balance of speed, quality, and cost. Model selector showing available models and multipliers
ModelProviderMultiplierPlan Requirements
AutoZencoder1×All plans
Auto+Zencoder2.5×All plans
Sonnet 4Anthropic2×All plans
Haiku 4.5 Parallel ThinkingAnthropic1×Starter, Core, Advanced, Max
Sonnet 4 Parallel ThinkingAnthropic3×Starter, Core, Advanced, Max
Sonnet 4.5 Parallel Thinking***Anthropic3×Starter, Core, Advanced, Max
Opus 4.1**Anthropic10×Advanced, Max
Opus 4.1 Parallel Thinking**Anthropic15×Advanced, Max
Gemini 2.5 ProGoogle1×All plans
GPT-5*OpenAI1×Starter, Core, Advanced, Max
GPT-5 Codex*OpenAI1×Starter, Core, Advanced, Max
Grok Code Fast 1xAI0.25×All plans
***Sonnet 4.5 Parallel Thinking has been available in Zencoder since September 29, 2025 - from day zero when Anthropic released it. This is the first model we’ve tested that feels truly purpose-built for spec-driven development, with persistent state tracking, unparalleled parallel execution, and high verification bias ensuring accuracy and reliability in complex coding tasks.*GPT-5 support has been available in Zencoder since August 8 - less than 24 hours after its public release. GPT-5 Codex is optimized for code generation tasks.
Grok Code Fast 1 at 0.25× multiplier is our most cost-efficient model offering, followed by Auto and Gemini 2.5 Pro at 1×. All multipliers and model availability are subject to change. Model availability evolves over time. We may add, remove, or upgrade models without notice as providers change capabilities and pricing. Your selector will always reflect the currently supported set for your plan.

​
Cost multipliers and Premium LLM calls

Zencoder measures usage in Premium LLM calls. Some models use a multiplier to reflect provider costs or parallel reasoning. See the table above and the screenshot for current multipliers.
Example: If a response would normally consume 10 Premium LLM calls, running it with Opus 4.1 (10×) would count as 100 calls. For details about daily limits and accounting, see Pricing & Plans.

​
Plan-based availability

Not all plans have access to all models. The selector shows what your current plan can use. If you recently upgraded but don’t see expected options, restart your IDE to refresh entitlements.
  • Starter, Core, Advanced and Max plans: Model model selector with allocated models, including GPT-5 access
  • Legacy plans: Model selector is not available

​
Bring Your Own Key (BYOK)

You can use your own API key for certain providers. This is useful if you:
  • Want to remove daily Premium LLM call limits for those requests
  • Prefer billing directly with the provider
Supported providers in the UI today: Anthropic and OpenAI.
BYOK is currently available on all plans, including Free, Starter, Core, Advanced, and Max.

​
Enable BYOK

1

Open Zencoder menu

Click the Zencoder icon in your IDE, then click the ⋯ (three dots) menu.
2

Choose User API key

Select User API key from the dropdown to open Custom API keys.
3

Enter keys and use

Paste your Anthropic or OpenAI key, save, then pick a matching provider model in the model selector.Custom API keys screen with Anthropic and OpenAI fields
When BYOK is active for a given provider/model, those requests can bypass Zencoder’s daily Premium LLM call limits - usage and billing are governed by your provider account. Make sure you understand the provider’s pricing before enabling heavy workloads.
Keep your keys secure. Only paste keys into the official Zencoder UI. If you rotate/regenerate keys at the provider, update them in Zencoder as well.

​
When to switch models

  • Use the Auto model for most coding tasks - it balances speed, quality, and cost
  • Use the Auto+ for superior performance on complex tasks (2.5× multiplier)
  • Use the Haiku 4.5 Parallel Thinking if you want speed and cost-efficient option (1x multiplier)
  • Use Grok Code Fast 1 when you need the most cost-efficient option (0.25× multiplier)
  • Use Gemini 2.5 Pro for a good balance of cost-efficiency and capability (1× multiplier)
  • Use Sonnet 4.5 Parallel Thinking for spec-driven development tasks requiring persistent state tracking and parallel execution
  • Use Parallel/advanced reasoning models (e.g., Sonnet 4 Parallel Thinking, Opus 4.1 Parallel Thinking) for complex planning or intricate refactors
  • Use GPT-5 Codex for specialized code generation tasks
  • Use Opus 4.1 models (Advanced and Max plans) for the most challenging tasks requiring maximum capability, keeping the higher multiplier in mind
Heavier reasoning models may produce better results on difficult tasks but at higher call consumption. Keep separate chats for different experiments to maintain clarity.

​
Troubleshooting

Make sure you’re on the new plans. Sign out/in to refresh entitlements.
Restart your IDE after signing back in.
Confirm you selected a model that matches the provider whose key you added, and that the key is valid/active.

Pricing & Plans

Learn how Premium LLM calls and daily limits work

Coding Agent

See how models are applied in the full agentic workflow