Skip to content
Tier B — Production
Runs in:CNMade in:China
Z.ai (GLM / Zhipu)

CogView-4

Tier B — Production

Tokonomix Editorial Team·Reviewed by Mes Kalkan··
Section 01

Speed analysis

Latency measured across all benchmark runs. P50 (median) and P95 (95th percentile) give a realistic picture of response speed under normal and peak load.

P50 latency (median)P95 latency53 runs
6686978512884159821908107-0907-22ms
Section 02

Quality scores

Evaluation results from judge-model scoring across diverse task categories. Scores reflect coherence, accuracy and instruction-following.

0
Coding
0
Creative
Section 03

Tokens per second

Throughput in tokens per second, derived from measured P50 latency. Higher is better; fluctuations track provider-side load.

Throughput (tokens / s)25 / avg 25
309

Estimated from P50 latency × 200 output tokens — the absolute number depends on this assumption; the trend is what matters.

Section 04

Capabilities

note: plain "cogview-4" = Unknown Model; use dated slugsource: zaiimage generation
Section 05

Availability

Availability

No measurements yet

We haven't recorded enough API calls to show availability stats for this model. Data appears once the model starts receiving live traffic.

Section 06

Tokonomix benchmark verdicts

⚖️
Endorsed by 1 judge
Independent LLM judges evaluated this model on our weekly intelligence tests
claude-sonnet-4-50/100 · 4 runs
0 correct0 partial4 wrong0% accuracy
2026-07-19

CogView-4 maintains steady performance across image generation benchmarks

CogView-4 continues to demonstrate consistent performance in the current benchmark window, maintaining the same capability profile as its debut. The model shows no significant changes across key image generation metrics, suggesting a stable release without major updates during this period. Users can expect reliable image generation functionality that remains competitive within its tier. The model's performance characteristics have remained unchanged, indicating a focus on stability rather than rapid iteration. This consistency may benefit users seeking predictable outputs for production workflows. For those evaluating CogView-4, the current benchmarks confirm that its strengths and limitations from the previous window remain intact. The lack of benchmark fluctuations suggests either model stability or minimal real-world usage variation during this measurement period. Organizations already using CogView-4 should not expect different performance characteristics from earlier deployments. The steady state of capabilities means CogView-4 continues to serve its original use cases without expansion into new modalities or significant performance improvements. Users seeking cutting-edge advances may want to monitor future benchmark windows for potential updates, while those prioritizing consistency will find the current stability reassuring.

Quality

Latency p50

Test runs

0

Performance remains stable Consistent image generation quality
Last automated test
Jul 22, 2026 · 08:02 UTC · Speed benchmark
P50 latency
7848 ms
P95 latency
8000 ms
Errors
0 / 6 runs
Last reviewed by Tokonomix Team·July 22, 2026