vs
DeepSeek vs Inception
DeepSeek serves MIT-licensed open models with 1M context at low first-party prices. Inception serves closed diffusion LLMs built for speed.
By The Subconscious Team · Updated
DeepSeek vs Inception: key differences
DeepSeek's API offers V4.1 Flash at $0.30 in and $1.20 out at peak, half price off-peak, with 1M context, 384K max output and very cheap cache hits. V4.1 Flash runs around 209 tokens per second there, per Artificial Analysis, and data on the hosted API is stored in China. Inception's Mercury 2.5 costs $0.20 in and $0.75 out with 260K context and tool use. Inception reports over 1,100 tokens per second, and Artificial Analysis measured about 640.
DeepSeek fits cost-sensitive agents and batch work that can wait for off-peak hours, plus teams that want open weights to self-host or fine-tune. Inception fits work where speed is the product, like autocomplete and voice, and its Bedrock and Azure listings help teams that cannot use a China-hosted API.
What DeepSeek and Inception do
DeepSeek
DeepSeek is the Chinese lab whose open-weight models reset price expectations for the whole market. Its main model is V4.1 Flash, with 1M context and 384K max output, which shipped September 10, 2026 with built-in image understanding at $0.30 in and $1.20 out at peak. DeepSeek is phasing out V4 Pro: since September 14, deepseek-v4-pro requests route to V4.1 Flash at Flash rates until V4.1 Pro launches. Cache hits cost a few cents per million or less, and the weights ship on Hugging Face under an MIT license.
Example models: DeepSeek V4.1 Flash
Full DeepSeek profile
Inception
Inception builds diffusion large language models (dLLMs). Every other LLM on this list generates text one token at a time; Inception's Mercury models start from a rough draft of a whole block and refine it over a few passes, producing many tokens in parallel. That ties speed less to output length and lets the models run fast on ordinary NVIDIA GPUs rather than custom chips. The company was founded by professors from Stanford, UCLA and Cornell behind core work on diffusion, Flash Attention and DPO, and is led by Stanford's Stefano Ermon. It raised a $50M seed in November 2025 led by Menlo Ventures, with Mayfield, M12, NVentures, Snowflake and Databricks, plus Andrew Ng and Andrej Karpathy as angels.
Example models: Mercury 2.5, Mercury 2
Full Inception profileShould you choose DeepSeek or Inception?
DeepSeek vs Inception at a glance
| Attribute | DeepSeek |  Inception |
|---|---|---|
| Model access | Open weights (MIT) | Closed weights (Mercury dLLMs) |
| Flagship models | DeepSeek V4.1 Flash | Mercury 2.5, Mercury 2 |
| Speed | ~209 tok/s on V4.1 Flash | 1,100+ tok/s (vendor); ~640 measured |
| Price | Off-peak hours at half price | $0.20 in, $0.75 out (Mercury 2.5) |
| Customization | Open weights to fine-tune | Fine-tuning on request |
| Deployment | First-party API, Hugging Face weights | Serverless API, Bedrock, Azure, private |
| Long context | 1M, 384K max output | 260K |
Frequently asked questions
What is the difference between DeepSeek and Inception?
DeepSeek serves MIT-licensed open models with 1M context at low first-party prices. Inception serves closed diffusion LLMs built for speed.
When should I choose DeepSeek over Inception?
Open weights to self-host or fine-tune; Long-context agents with cheap cache hits; Batch work in off-peak windows.
When should I choose Inception over DeepSeek?
Autocomplete and fast code edits; Voice agents and real-time chat; Buying through Bedrock or Azure.
Is DeepSeek or Inception cheaper?
DeepSeek: Off-peak hours at half price. Inception: $0.20 in, $0.75 out (Mercury 2.5). The cheaper choice depends on the model and workload.
Which has more context, DeepSeek or Inception?
DeepSeek: 1M, 384K max output. Inception: 260K.
Related comparisons
Running long-horizon agents?
If your agents run past 200K tokens, compare both against Subconscious. Our inference stack treats a long-horizon trace as the primary workload, so speed, cost, and accuracy hold up deep into the trace.