Maximum throughput
Up to 200 tokens/s for latency-sensitive workloads: real-time chat, agentic loops, IDE copilots, and high-concurrency production.
200 tok/s on the same frontier intelligence
GLM-5.3-FlashX is the high-speed variant of GLM-5.3-Flash, officially launched by Z.ai on September 18, 2026. Previously previewed under the code name Ox Alpha, GLM-5.3-Flash has seen rapid adoption among developers worldwide. Facing surging demand, Z.ai invested further in Infra and inference optimization, backed by 100,000 domestic chips, to launch FlashX at up to 200 tokens/s.
Same intelligence
FlashX retains the full reasoning, coding, and multimodal capabilities of GLM-5.3-Flash — no quality trade-off.
Aligned pricing
Same usage-based pricing as GLM-5.3-Flash: $0.14/M input, $0.52/M output. Speed without the premium.
GLM-5.3-FlashX vs GLM-5.3-Flash
FlashX is the faster variant. Both share the same architecture and intelligence. Choose based on your latency requirements.
| Dimension | GLM-5.3-Flash | GLM-5.3-FlashX |
|---|---|---|
| Speed | Standard | Up to 200 tok/s |
| Intelligence | Frontier | Frontier (same) |
| Architecture | 320B total / 18B active MoE | 320B total / 18B active MoE |
| Context | ~1.3M tokens | ~1.3M tokens |
| Modalities | Text · Image · Video → text | Text · Image · Video → text |
| Pricing (input) | $0.14 / 1M tokens | $0.14 / 1M tokens |
| Pricing (output) | $0.52 / 1M tokens | $0.52 / 1M tokens |
| Infrastructure | Standard inference | 100K domestic chips + optimized inference |
Both models are available
GLM-5.3-FlashX does not replace GLM-5.3-Flash. GLM-5.3-Flash remains available for workloads where standard speed is sufficient. Confirm current pricing and rate limits for both models in the live TokenRa console.
Where GLM-5.3-FlashX fits
FlashX is optimized for workloads where throughput and latency are critical, without compromising on intelligence.
- Real-time chat and conversational agents that require sub-second response times.
- Agentic loops with multiple sequential tool calls where cumulative latency matters.
- IDE copilots and inline code completions that demand instant feedback.
- High-concurrency production serving with strict latency budgets.
- Long-context reasoning where output token generation speed dominates end-to-end latency.
How to get a GLM-5.3-FlashX API key
Register for TokenRa, open the dashboard, create an API key, confirm that GLM-5.3-FlashX is enabled for your account, and then send an OpenAI-compatible request from your server. Keep the key private and verify the live model identifier before production use.
Create a TokenRa account
Register, then open the TokenRa dashboard.
Create your API key
Open the API key section, create a key, and store it server-side.
Send a test request
Confirm the enabled model ID glm-5.3-flashx, test a representative workload, and add production controls.
curl https://tokenra.io/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"glm-5.3-flashx","messages":[{"role":"user","content":"Write a Python function to sort a list"}]}'GLM-5.3-FlashX pricing
GLM-5.3-FlashX pricing is aligned with GLM-5.3-Flash: $0.14 per 1M input tokens, $0.52 per 1M output tokens, and $0.04 per 1M cached read tokens through TokenRa. No credit card is required to get started. Confirm current rates in the live TokenRa console before production use.
$0.14 / 1M tokens
Standard input pricing for GLM-5.3-FlashX.
$0.52 / 1M tokens
Generated output pricing for GLM-5.3-FlashX.
Account-based
Check the TokenRa console for current RPM and quota limits.
Z.ai (Zhipu AI)
GLM-5.3-FlashX is developed and operated by Z.ai (Zhipu AI). It was previously previewed under the code name Ox Alpha. TokenRa provides API access as a routing gateway, and is not the model developer or owner.
Data handling
Prompts and completions are retained by the provider and are not used for training. Review the applicable provider and TokenRa terms before sending sensitive or regulated information.
Production checklist
- Keep API keys on the server and separate credentials by environment.
- Confirm the 200 tok/s throughput and supported input modalities against the live integration before relying on them.
- Run representative coding, agentic, reasoning, and visual-context workloads.
- Set explicit timeouts, bounded retries, tracing, and application-side budgets.
- Review data handling, provider terms, privacy requirements, and output rights for your use case.
GLM-5.3-FlashX FAQ
What is GLM-5.3-FlashX?
GLM-5.3-FlashX is the high-speed variant of GLM-5.3-Flash, officially launched by Z.ai on September 18, 2026. It delivers up to 200 tokens/s while retaining the same frontier intelligence.
How fast is GLM-5.3-FlashX compared to GLM-5.3-Flash?
FlashX reaches up to 200 tokens/s, making it ideal for real-time and high-concurrency workloads. GLM-5.3-Flash runs at standard speed with the same intelligence.
Is GLM-5.3-FlashX more expensive?
No. Pricing is aligned with GLM-5.3-Flash: $0.14/M input and $0.52/M output tokens. Check the live console for current rates.
How do I get an API key?
Register for TokenRa, create an API key, and use the model identifier glm-5.3-flashx in your requests.
Does GLM-5.3-FlashX support the same input types?
Yes — text, image, and video input (to text output), subject to the enabled integration.
Who develops GLM-5.3-FlashX?
Z.ai (Zhipu AI). TokenRa is an API routing gateway and is not the model developer.
GLM-5.3-FlashX is developed by Z.ai (Zhipu AI). The provider and related names belong to their respective rights holders. TokenRa provides API access as a routing gateway and is not the model developer or owner.