vs
Relace vs Inception
Relace trains small tool models for coding agents. Inception sells general diffusion reasoning models built for fast output.
By The Subconscious Team · Updated
Relace vs Inception: key differences
Relace's relace-apply-3 merges lazy edit snippets into files at about 10,000 tokens per second with up to 128K tokens, and its toolkit adds agentic codebase search and a compaction model at 50,000 tokens per second. It is hosted or self-hosted. Inception's Mercury 2.5 is a general reasoning model with tool use, JSON mode and 260K context at $0.20 in and $0.75 out. Inception reports over 1,100 tokens per second; Artificial Analysis measures about 640.
Relace takes utility work off a main model and does not serve general models. Mercury can be the main model for fast loops like code autocomplete and edits, though it trails frontier models on hard reasoning and long agent tasks. A coding product could pair them.
What Relace and Inception do
Relace
Relace trains small, fast models that act as tools for coding agents. Its best-known product is Instant Apply: a frontier model writes a lazy edit snippet, and relace-apply-3 merges it into the original file at about 10,000 tokens per second with 128K tokens of input and output. Relace says this runs over 3x faster and cheaper than having the big model rewrite the file. It exposes both a REST endpoint and an OpenAI-compatible one, and the model is also listed on OpenRouter.
Example models: relace-apply-3, Relace agentic search
Full Relace 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 Relace or Inception?
Relace vs Inception at a glance
| Attribute | Relace |  Inception |
|---|---|---|
| Model access | Specialist models | Closed weights (Mercury dLLMs) |
| Flagship models | relace-apply-3, agentic search | Mercury 2.5, Mercury 2 |
| Speed | ~10,000 tok/s apply | 1,100+ tok/s (vendor); ~640 measured |
| Price | 3x+ cheaper than full rewrites | $0.20 in, $0.75 out (Mercury 2.5) |
| Customization | Unknown | Fine-tuning on request |
| Deployment | Hosted API or self-hosted | Serverless API, Bedrock, Azure, private |
| Long context | 128K max | 260K |
Frequently asked questions
What is the difference between Relace and Inception?
Relace trains small tool models for coding agents. Inception sells general diffusion reasoning models built for fast output.
When should I choose Relace over Inception?
Applying AI edits to user codebases; Fast search across large repos; Self-hosted coding tools.
When should I choose Inception over Relace?
A fast general model for real-time loops; Code autocomplete and quick edits; Cheap reasoning with tool use and 260K context.
Is Relace or Inception cheaper?
Relace: 3x+ cheaper than full rewrites. Inception: $0.20 in, $0.75 out (Mercury 2.5). The cheaper choice depends on the model and workload.
Which has more context, Relace or Inception?
Relace: 128K 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.