vs
Groq vs Inception
Both sell speed. Groq gets it from custom LPU chips running open models. Inception gets it from diffusion models running on standard GPUs.
By The Subconscious Team · Updated
Groq vs Inception: key differences
Groq serves open models like GPT-OSS 120B and Qwen 3.8 27B at 500 to 1,000 tokens per second with tight tail latency, capped around 131K context and with no fine-tuned model hosting. Inception's Mercury models generate many tokens in parallel instead of one at a time. Mercury 2.5 is a closed reasoning model with tool use and 260K context at $0.20 in and $0.75 out. Inception reports 1,100+ tokens per second; Artificial Analysis measures about 640.
Groq offers open weights and published per-model speeds on a shrinking catalog. Inception offers longer context, fine-tuning and private deployments on request, and availability on Bedrock and Azure, but only for its own models. Teams building voice or autocomplete should test both on their own prompts.
What Groq and Inception do
Groq
Groq serves open models on its own chip, the LPU, which keeps model weights in on-chip SRAM and runs a deterministic schedule instead of waiting on GPU memory. Groq publishes 1,000 tokens per second on GPT-OSS 20B and 500 on GPT-OSS 120B. Latency stays tight between median and tail, which matters for strict SLAs. The API is OpenAI-compatible and also hosts Whisper for speech to text plus an agentic system called Groq Compound with built-in search and code execution.
Example models: GPT-OSS 120B, Qwen 3.8 27B
Full Groq 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 Groq or Inception?
Groq vs Inception at a glance
| Attribute | Groq |  Inception |
|---|---|---|
| Model access | Open weights | Closed weights (Mercury dLLMs) |
| Flagship models | GPT-OSS 120B, Qwen 3.8 27B | Mercury 2.5, Mercury 2 |
| Speed | 500–1,000 tok/s | 1,100+ tok/s (vendor); ~640 measured |
| Price | Near the floor on small models | $0.20 in, $0.75 out (Mercury 2.5) |
| Customization | No fine-tuned model hosting | Fine-tuning on request |
| Deployment | GroqCloud API | Serverless API, Bedrock, Azure, private |
| Long context | Around 131K max | 260K |
Frequently asked questions
What is the difference between Groq and Inception?
Both sell speed. Groq gets it from custom LPU chips running open models. Inception gets it from diffusion models running on standard GPUs.
When should I choose Groq over Inception?
Fast output on open-weight models; Tight tail latency; Low prices on small models.
When should I choose Inception over Groq?
Fast output with 260K context; Buying through Bedrock or Azure; Fine-tuning or private deployment on request.
Is Groq or Inception cheaper?
Groq: Near the floor on small models. Inception: $0.20 in, $0.75 out (Mercury 2.5). The cheaper choice depends on the model and workload.
Which has more context, Groq or Inception?
Groq: Around 131K max. 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.