AI Provider Setup · BYOK

Connect DeepSeek

Cheapest reasoning per token — strong chain-of-thought at a fraction of Claude's price.

DeepSeek's reasoning models deliver solid multi-step thinking at roughly 5–10x lower cost than comparable Claude or GPT tiers. Good fit for bulk analytical workloads where you'd otherwise burn through budget. The trade-off: slower time-to-first-token than the US-hosted frontier labs.

Best used for

Key details

Where to generate your key
https://platform.deepseek.com/api_keys
Expected key format
Starts with sk-
Environment variable name (if self-hosting)
DEEPSEEK_API_KEY
Provider pricing
https://api-docs.deepseek.com/quick_start/pricing
Official docs
DeepSeek API docs

Step-by-step setup

  1. 1 Create a DeepSeek platform account

    Go to platform.deepseek.com and sign up. International sign-up works with email — you don't need a Chinese phone number.

  2. 2 Top up credits

    DeepSeek is prepaid. Open Billing in the left sidebar and top up. $5 goes a long way given the price per token.

  3. 3 Open the API Keys page

    Navigate to platform.deepseek.com/api_keys. You'll see an empty list until you create one.

  4. 4 Create a key labeled Admaxxer

    Click Create API Key, name it Admaxxer, and create. The key starts with sk- and is shown only once — copy it immediately.

  5. 5 Paste into Admaxxer

    Go to Settings then AI Providers, find the DeepSeek row, paste the key, and click Connect.

  6. 6 Pick a default model

    deepseek-reasoner has the full chain-of-thought. deepseek-chat is faster for conversational turns without explicit reasoning traces.

  7. 7 If the connection test fails

    Confirm the key starts with sk-, credits are loaded, and the key hasn't been revoked. DeepSeek's status page is at status.deepseek.com.

Test the connection inside Admaxxer

Once you've pasted your key into /settings/ai-providers, hit Test. Admaxxer makes a single no-cost call against the provider's /v1/models (or equivalent) endpoint and reports back:

Available models

These are the pinned model IDs Admaxxer tests against. The model picker in chat will show these plus any live-catalog entries the provider returns.

Model ID Display name Context Tools Best for
deepseek-reasoner DeepSeek Reasoner (recommended) 128K Yes Full chain-of-thought reasoning at the lowest price on the market.
deepseek-chat DeepSeek Chat 128K Yes Faster, cheaper, no explicit CoT. Good for routing and short answers.
deepseek-coder DeepSeek Coder 128K Yes Code-specialized variant. Use when the chat task is mostly code-generation or code-review.

Common errors and fixes

401

When it happens: On Connect if the key is invalid or revoked.

Fix: Regenerate at platform.deepseek.com/api_keys and paste the new key.

402

When it happens: If your DeepSeek balance hits zero.

Fix: Top up at platform.deepseek.com/billing. DeepSeek does not auto-recharge.

429

When it happens: Under very heavy concurrent load.

Fix: DeepSeek's default rate limit is generous. If you hit it, space out parallel requests.

timeout

When it happens: If DeepSeek's region is under pressure — latency can spike.

Fix: Retry with exponential backoff. For critical paths, use OpenRouter to fail over.