# Inco AI launches Day-0 support for GLM 5.3

> Source: <https://inco.ai/blog/glm-5-3/>
> Published: 2026-08-28 00:00:00+00:00

# Inco AI launches Day-0 support for GLM 5.3

[Overview](#overview)

[GLM 5.3](https://z.ai/blog/glm-5.3) arrives today with day-0 serving support
from Inco AI on
[TokenRouter](https://www.tokenrouter.com/) compute. The
release pairs [Z.ai's](https://z.ai/) new model with our core inference
optimization technology:

- A
[DFlash 2 checkpoint](https://huggingface.co/incoai/GLM-5.3-DFlash2)for speculative decoding. - An
[NVFP4 checkpoint](https://huggingface.co/incoai/GLM-5.3-NVFP4)for efficient Blackwell inference with native accuracy. [Inco Engine](https://platform.inco.ai/playground?model=glm-5.3)for end-to-end inference performance.

As a day-0 launch partner, we are releasing both checkpoints alongside the
model endpoint. Together, the checkpoints and Inco
Engine deliver up to **4.4× throughput** versus the native
[FP8 checkpoint](https://huggingface.co/zai-org/GLM-5.3) with autoregressive
decoding at concurrency 1.

[DFlash 2 for GLM 5.3](#dflash-2-for-glm-53)

[DFlash 2](/blog/dflash2/) is a parallel drafter for speculative decoding: it
predicts candidate tokens in one pass and lets the target model verify them as
a block.

Acceptance length (AL) measures how many tokens each draft–verify cycle yields, including the verifier's next token. A longer accepted block means fewer full target-model passes for the same output. DFlash 2 improves AL by keeping the parallel draft design while selecting a more coherent path through each position's candidates.

Acceptance length is only half of the serving result: a drafter also adds work to each cycle. For GLM 5.3, we therefore report end-to-end decode throughput separately, comparing the model's native MTP path and DFlash 2 against autoregressive decoding. At concurrency 1, DFlash 2 reaches 383.3 output tok/s on MATH-500, 366.6 on GSM8K, and 363.7 on HumanEval.

[NVFP4 for Blackwell](#nvfp4-for-blackwell)

We are also releasing an
[NVFP4 checkpoint](https://huggingface.co/incoai/GLM-5.3-NVFP4) for efficient
Blackwell deployment, with accuracy matching the native
[FP8 checkpoint](https://huggingface.co/zai-org/GLM-5.3) across the evaluation
suite.

| Precision | GPQA Diamond | AIME 2025 | MATH-500 | HLE | AA-LCR |
|---|---|---|---|---|---|
| FP8 | 91.1 | 94.3 | 95.6 | 35.9 | 73.6 |
| NVFP4 | 91.2 | 95.1 | 95.2 | 35.2 | 73.0 |

[Get access](#get-access)

Try GLM 5.3 in the browser or connect through the Inco API.

### Included in preview

- Browser playground
- OpenAI and Anthropic API dialects
- Streaming responses
- 1M-token context window

### List pricing

per 1M tokens- Input
- $1.40
- Cached input
- $0.26
- Output
- $4.40

[Get checkpoints](#get-checkpoints)

[Compute for GLM 5.3](#compute-for-glm-53)

Compute for this launch came from
[TokenRouter](https://www.tokenrouter.com/). We trained the DFlash 2 drafter
and produced the NVFP4 checkpoint on its Blackwell clusters, and we serve the
day-0 endpoint from the same pool. You can reach the endpoint with an existing
TokenRouter key.

[The bottom line](#the-bottom-line)

Ten days ago, we called [DFlash 2](/blog/dflash2/) the first piece of an
end-to-end serving stack. Launches like this are what the rest of it is for.

GLM 5.3 shipped today, and the stack shipped with it: a DFlash 2 drafter, an NVFP4 checkpoint, and a live endpoint on Inco Engine. Together they serve the model at up to 4.4× the throughput of the native FP8 checkpoint with autoregressive decoding, measured at concurrency 1.

If you want your model served like this, fine-tunes included, write to us:
[contact@inco.ai](mailto:contact@inco.ai).

[Get updates](#get-updates)

One email when we ship something new.

We will never share your email address.