vs
Z.ai vs Inception
Z.ai makes open-weight GLM models with a cheap coding plan. Inception makes closed diffusion models built for speed.
By The Subconscious Team · Updated
Z.ai vs Inception: key differences
Z.ai's GLM-5.3 costs $1.40 in and $4.40 out with 1M context at about 80 tokens per second, GLM-5.3-Flash costs $0.075 in and $0.25 out, and the GLM Coding Plan starts at $18 a month. Servers sit mostly in China. Inception sells closed Mercury diffusion models that refine a block of tokens in parallel instead of one at a time. Mercury 2.5 is a reasoning model with tool use and 260K context at $0.20 in and $0.75 out per million tokens; Inception reports over 1,100 tokens per second, and Artificial Analysis measures about 640.
Pick Z.ai for stronger open models, weights you can self-host, and cheap Claude Code use. Pick Inception when output speed is the main constraint and US cloud listings on Bedrock or Azure matter.
What Z.ai and Inception do
Z.ai
Z.AI is the international brand of Chinese lab Zhipu AI, maker of the GLM models. Its current flagship, GLM-5.3, shipped August 17, 2026 at $1.40 in and $4.40 out per million tokens, with cached input at $0.26. Its weights came out August 28 under a custom GLM-5.3 License that requires a security review for hosts with more than $10B in yearly revenue. GLM-5.3-Flash costs $0.075 in and $0.25 out, and several older Flash models are priced at zero, a real free tier instead of trial credits. GLM-5, released in February 2026, is a 744B mixture-of-experts model under an MIT license, and at launch it ranked first among open-weight models on the Artificial Analysis index with a record-low hallucination score.
Example models: GLM-5.3, GLM-5.3-Flash
Full Z.ai 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 Z.ai or Inception?
Z.ai vs Inception at a glance
| Attribute | Z.ai |  Inception |
|---|---|---|
| Model access | Open weights (GLM-5.3 License; MIT on Flash and older) | Closed weights (Mercury dLLMs) |
| Flagship models | GLM-5.3, GLM-5.3-Flash | Mercury 2.5, Mercury 2 |
| Speed | ~80 tok/s on GLM-5.3 | 1,100+ tok/s (vendor); ~640 measured |
| Price | $1.40 in, $4.40 out (GLM-5.3); free Flash tier | $0.20 in, $0.75 out (Mercury 2.5) |
| Customization | Open weights; GLM-5.3 License adds review for large hosts | Fine-tuning on request |
| Deployment | API, GLM Coding Plan | Serverless API, Bedrock, Azure, private |
| Long context | 1M (GLM-5.3) | 260K |
Frequently asked questions
What is the difference between Z.ai and Inception?
Z.ai makes open-weight GLM models with a cheap coding plan. Inception makes closed diffusion models built for speed.
When should I choose Z.ai over Inception?
Open weights to self-host; Cheap Claude Code use via the Coding Plan; 1M context on GLM-5.3.
When should I choose Inception over Z.ai?
Very fast output on standard GPUs; Cheap reasoning with tool use and 260K context; Buying through AWS Bedrock or Azure.
Is Z.ai or Inception cheaper?
Z.ai: $1.40 in, $4.40 out (GLM-5.3); free Flash tier. Inception: $0.20 in, $0.75 out (Mercury 2.5). The cheaper choice depends on the model and workload.
Which has more context, Z.ai or Inception?
Z.ai: 1M (GLM-5.3). 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.