GLM-5.3-Flash: Benchmarks, Pricing & Context Window
GLM-5.3-Flash is a language model from ZAI, released in August 2026, with multimodal input, a 1.0M-token context window, and pricing from $0.150/M input, $0.030/M cached input, $0.500/M output.
GLM-5.3-Flash is the first natively multimodal model in Z.AI's GLM-5 series: a 320B-total / 18B-active MoE with hybrid sparse and linear attention (plus mHC), a 1M-token context window, and strong coding and long-horizon agent performance.
GLM-5.3-Flash benchmarks
Capability tiers
Standing within each category, adjusted for leaderboard depth.
Real tasks performance
High-confidence performance for GLM-5.3-Flash across real-world prompt categories. Only 95% intervals at most 4 points wide are shown.
Performance by conversation depth
How GLM-5.3-Flash holds up as conversations get longer.
Quality Tracker
GLM-5.3-Flash Performance Across Datasets
Scores sourced from the model's scorecard, paper, or official blog posts
llm-stats.com - Wed Sep 09 2026GLM-5.3-Flash pricing
Providers
GLM-5.3-Flash starts at $0.150 per million input tokens and $0.500 per million output tokens via DeepInfra. Reused prompt prefixes cost $0.0300 per million cached input tokens. See all 4 providers below with their per-token pricing, latency, throughput, and modality support.
| Provider | Input $/M | Cached input $/M | Output $/M | Context in / out | TTFT p95 s | Output p5 c/s | Modalities in / out |
|---|---|---|---|---|---|---|---|
 DeepInfra | $0.150 | $0.0300 | $0.500 | 1.0M/1.0M | 2.96 | 27 | / |
 FriendliAI | $0.150 | $0.0300 | $0.500 | 1.0M/131.1K | 3.21 | 20 | / |
 Novita | $0.150 | $0.0300 | $0.500 | 1.0M/131.1K | 10.88 | 10 | / |
 ZAI | $0.150 | $0.0300 | $0.500 | 1.0M/131.1K | 8.96 | 6 | / |
Cached input is the discounted price for prompt tokens served from a provider cache. TTFT is time to first token. Output is characters per second; p5 is the sustained floor exceeded by 95% of observed requests.
GLM-5.3-Flash model size
GLM-5.3-Flash has 320 billion parameters and was trained on 30 trillion tokens. See how it compares to other models in the same parameter range.
GLM-5.3-Flash context window
Input and output token limits for GLM-5.3-Flash, plus how it ranks on long-context understanding.
Try now
huggleMake it with
GLM-5.3-Flash.
GLM-5.3-Flash latency
GLM-5.3-Flash time to first token, sustained output throughput, and failed-request rate from live model usage over the trailing 7 days.
Provider operational metrics
Time to first token, output throughput, and failed-request rate from live model usage
GLM-5.3-Flash examples
Recent arena outputs from GLM-5.3-Flash, picked from the highest-ranked matchups.
GLM-5.3-Flash license
GLM-5.3-Flash is released under the MIT license, which permits commercial use, has 320.0B parameters.
- License
- MIT
- Commercial use allowed
- Parameters
- 320.0B
MIT License - allows commercial use
GLM-5.3-Flash resources
Official sources for GLM-5.3-Flash: provider documentation, official playground, paper or system card, official launch post, source repository, model weights.
GLM-5.3-Flash vs other models
The most-compared alternatives to GLM-5.3-Flash are Claude Opus 4.8, Muse Spark 1.1, Claude Opus 4.7. Open any pair side-by-side for benchmarks, pricing, context, and latency.
Models like GLM-5.3-Flash
Models ranked just above and below GLM-5.3-Flash by LLM Stats score.
FAQ
Common questions about GLM-5.3-Flash.